{
  "snapshot": "2026-09-30",
  "compiled_at": "2026-10-01T03:08:44.747044+00:00",
  "source_sha256": "7c9acec74c53f350157b803fb83b103af1edbfcd6f7c4a2e9237dd6f1c383ca5",
  "scope": "129 recovered benchmark and service-probe rows, not a complete history of every experiment. Public metrics only; private paths, endpoints, raw prompts and case records omitted.",
  "definitions": {
    "wall_s": "Sum of recorded client request elapsed times, excluding between-request/startup delays.",
    "rates": "Server-reported tokens/sec; distribution over requests, not aggregate token/time throughput.",
    "quality": "Original pinned public123 and frozen24 scorers. Do not compare percentages across suites.",
    "p95": "Nearest rank, unweighted per recorded case.",
    "missing": "Null/recorded-case counts denote absent telemetry; missing fields are not evidence of zero failures/retries/truncations."
  },
  "hardware": [
    {
      "name": "Xeon · dual RTX 4060 Ti",
      "cpu": "Intel Xeon Gold 6242R · 20 cores / 40 threads",
      "platform": "HP Z8 G4",
      "memory": "256 GB DDR4",
      "installed_gpus": "2 × NVIDIA RTX 4060 Ti · 16 GB each",
      "storage": "HDD-backed model and working storage",
      "os": "Not captured in the recovered hardware summary"
    },
    {
      "name": "Ryzen · dual RTX 4060 Ti",
      "cpu": "AMD Ryzen 5 9600X · 6 cores / 12 threads",
      "platform": "ASRock X870E Taichi",
      "memory": "32 GB DDR5",
      "installed_gpus": "2 × NVIDIA RTX 4060 Ti · 16 GB each",
      "storage": "2 × 500 GB NVMe: OS / Docker / data and dedicated read-only model weights",
      "os": "Ubuntu Server 26.04.1 LTS"
    },
    {
      "name": "Core i5 · Arc Pro B70",
      "cpu": "Intel Core i5-11600K · 6 cores / 12 threads",
      "platform": "MSI MPG Z590 GAMING PLUS (MS-7D07)",
      "memory": "128 GB DDR4 · 4 × 32 GB DIMMs · 125.6 GiB OS-visible",
      "installed_gpus": "1 × Intel Arc Pro B70 · 32 GB (benchmark accelerator)",
      "storage": "Samsung 990 EVO Plus 1 TB NVMe + Samsung 990 PRO 2 TB NVMe; 12 TB SATA HDD",
      "os": "Ubuntu 26.04.1 LTS"
    },
    {
      "name": "Core i5 · Arc A770",
      "cpu": "Intel Core i5-12600K · 10 cores / 16 threads",
      "platform": "ASRock H670M Pro RS",
      "memory": "32 GB DDR4-3200 · 2 × 16 GB DIMMs · 31.1 GiB OS-visible",
      "installed_gpus": "1 × Intel Arc A770 · 16 GB (benchmark accelerator)",
      "storage": "Samsung 990 EVO Plus 1 TB NVMe; 12 TB SATA HDD; 4 TB USB disk",
      "os": "Debian GNU/Linux 13 (trixie)"
    },
    {
      "name": "Core Ultra · RTX 5090",
      "cpu": "Intel Core Ultra 9 285K · 24 threads; core count not captured in this inventory",
      "platform": "Motherboard not captured in the recovered records",
      "memory": "64 GB installed · 63.5 GiB OS-visible; memory type not recorded",
      "installed_gpus": "1 × NVIDIA RTX 5090 · 32 GB (benchmark accelerator)",
      "storage": "Root volume: 1,906 GiB; drive model / media type not captured",
      "os": "Microsoft Windows 11 Home"
    }
  ],
  "runs": [
    {
      "id": "run-1",
      "run": "20260930T033730Z-cpu40/flash.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "IK llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": null,
      "decode": 14.85,
      "ttft": null,
      "latency": 184.85000000000002,
      "wall": 369.70000000000005,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 16384,
      "slots": 1,
      "kv": null,
      "date": "20260930T033730Z",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 9,
        "checks_possible": 9,
        "check_percent": 100,
        "wall_s": 369.70000000000005,
        "median_decode_tps": 14.85,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 29.7,
            "min": 14.6,
            "median": 14.85,
            "mean": 14.85,
            "p95": 15.1,
            "max": 15.1
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 2,
            "total": 369.70000000000005,
            "min": 150.4,
            "median": 184.85000000000002,
            "mean": 184.85000000000002,
            "p95": 219.3,
            "max": 219.3
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 2,
            "total": 3020,
            "min": 1507,
            "median": 1510,
            "mean": 1510,
            "p95": 1513,
            "max": 1513
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 4,
          "checks_possible": 4,
          "check_percent": 100,
          "wall_s": 219.3,
          "median_decode_tps": 14.6,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.6,
              "min": 14.6,
              "median": 14.6,
              "mean": 14.6,
              "p95": 14.6,
              "max": 14.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 219.3,
              "min": 219.3,
              "median": 219.3,
              "mean": 219.3,
              "p95": 219.3,
              "max": 219.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1507,
              "min": 1507,
              "median": 1507,
              "mean": 1507,
              "p95": 1507,
              "max": 1507
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 150.4,
          "median_decode_tps": 15.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.1,
              "min": 15.1,
              "median": 15.1,
              "mean": 15.1,
              "p95": 15.1,
              "max": 15.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 150.4,
              "min": 150.4,
              "median": 150.4,
              "mean": 150.4,
              "p95": 150.4,
              "max": 150.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1513,
              "min": 1513,
              "median": 1513,
              "mean": 1513,
              "p95": 1513,
              "max": 1513
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-2",
      "run": "20260930T034453Z-cpu36/flash.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "IK llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": null,
      "decode": 20.1,
      "ttft": null,
      "latency": 118.85,
      "wall": 237.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 16384,
      "slots": 1,
      "kv": null,
      "date": "20260930T034453Z",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 237.7,
        "median_decode_tps": 20.1,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 40.2,
            "min": 20,
            "median": 20.1,
            "mean": 20.1,
            "p95": 20.2,
            "max": 20.2
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 2,
            "total": 237.7,
            "min": 96.2,
            "median": 118.85,
            "mean": 118.85,
            "p95": 141.5,
            "max": 141.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 2,
            "total": 2801,
            "min": 1217,
            "median": 1400.5,
            "mean": 1400.5,
            "p95": 1584,
            "max": 1584
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 141.5,
          "median_decode_tps": 20,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 20,
              "min": 20,
              "median": 20,
              "mean": 20,
              "p95": 20,
              "max": 20
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 141.5,
              "min": 141.5,
              "median": 141.5,
              "mean": 141.5,
              "p95": 141.5,
              "max": 141.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1584,
              "min": 1584,
              "median": 1584,
              "mean": 1584,
              "p95": 1584,
              "max": 1584
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 96.2,
          "median_decode_tps": 20.2,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 20.2,
              "min": 20.2,
              "median": 20.2,
              "mean": 20.2,
              "p95": 20.2,
              "max": 20.2
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 96.2,
              "min": 96.2,
              "median": 96.2,
              "mean": 96.2,
              "p95": 96.2,
              "max": 96.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1217,
              "min": 1217,
              "median": 1217,
              "mean": 1217,
              "p95": 1217,
              "max": 1217
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-3",
      "run": "20260930T000829Z-smoke/27b.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "partial/probe",
      "cases": 1,
      "expected": 123,
      "success": 100,
      "quality": 100,
      "structure": 1,
      "strict": 1,
      "prefill": 602.710157558723,
      "decode": 51.74,
      "ttft": null,
      "latency": 5.774,
      "wall": 5.774,
      "requests": 1,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": "q4_0",
      "date": "20260930T000829Z",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": 1,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 5.774,
        "median_decode_tps": 51.74,
        "median_prefill_tps": 602.71,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 1,
        "request_count_recorded_cases": 1,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": 1,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 51.74,
            "min": 51.74,
            "median": 51.74,
            "mean": 51.74,
            "p95": 51.74,
            "max": 51.74
          },
          "prefill_tps": {
            "n": 1,
            "total": 602.710157558723,
            "min": 602.710157558723,
            "median": 602.710157558723,
            "mean": 602.710157558723,
            "p95": 602.710157558723,
            "max": 602.710157558723
          },
          "elapsed_s": {
            "n": 1,
            "total": 5.774,
            "min": 5.774,
            "median": 5.774,
            "mean": 5.774,
            "p95": 5.774,
            "max": 5.774
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 1,
            "total": 621,
            "min": 621,
            "median": 621,
            "mean": 621,
            "p95": 621,
            "max": 621
          },
          "completion_tokens": {
            "n": 1,
            "total": 246,
            "min": 246,
            "median": 246,
            "mean": 246,
            "p95": 246,
            "max": 246
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": 1,
          "checks_earned": 10,
          "checks_possible": 10,
          "check_percent": 100,
          "wall_s": 5.774,
          "median_decode_tps": 51.74,
          "median_prefill_tps": 602.71,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 1,
          "request_count_recorded_cases": 1,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": 1,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 51.74,
              "min": 51.74,
              "median": 51.74,
              "mean": 51.74,
              "p95": 51.74,
              "max": 51.74
            },
            "prefill_tps": {
              "n": 1,
              "total": 602.710157558723,
              "min": 602.710157558723,
              "median": 602.710157558723,
              "mean": 602.710157558723,
              "p95": 602.710157558723,
              "max": 602.710157558723
            },
            "elapsed_s": {
              "n": 1,
              "total": 5.774,
              "min": 5.774,
              "median": 5.774,
              "mean": 5.774,
              "p95": 5.774,
              "max": 5.774
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 621,
              "min": 621,
              "median": 621,
              "mean": 621,
              "p95": 621,
              "max": 621
            },
            "completion_tokens": {
              "n": 1,
              "total": 246,
              "min": 246,
              "median": 246,
              "mean": 246,
              "p95": 246,
              "max": 246
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-4",
      "run": "20260930T000829Z-smoke/9b.jsonl",
      "model": "Qwen3.5-9B · Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "partial/probe",
      "cases": 1,
      "expected": 123,
      "success": 100,
      "quality": 70,
      "structure": 1,
      "strict": 0,
      "prefill": 2061.855670103093,
      "decode": 46.76,
      "ttft": null,
      "latency": 3.901,
      "wall": 3.901,
      "requests": 1,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T000829Z",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": 0,
        "checks_earned": 7,
        "checks_possible": 10,
        "check_percent": 70,
        "wall_s": 3.901,
        "median_decode_tps": 46.76,
        "median_prefill_tps": 2061.856,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 1,
        "request_count_recorded_cases": 1,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": 1,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 46.76,
            "min": 46.76,
            "median": 46.76,
            "mean": 46.76,
            "p95": 46.76,
            "max": 46.76
          },
          "prefill_tps": {
            "n": 1,
            "total": 2061.855670103093,
            "min": 2061.855670103093,
            "median": 2061.855670103093,
            "mean": 2061.855670103093,
            "p95": 2061.855670103093,
            "max": 2061.855670103093
          },
          "elapsed_s": {
            "n": 1,
            "total": 3.901,
            "min": 3.901,
            "median": 3.901,
            "mean": 3.901,
            "p95": 3.901,
            "max": 3.901
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 1,
            "total": 621,
            "min": 621,
            "median": 621,
            "mean": 621,
            "p95": 621,
            "max": 621
          },
          "completion_tokens": {
            "n": 1,
            "total": 169,
            "min": 169,
            "median": 169,
            "mean": 169,
            "p95": 169,
            "max": 169
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": 0,
          "checks_earned": 7,
          "checks_possible": 10,
          "check_percent": 70,
          "wall_s": 3.901,
          "median_decode_tps": 46.76,
          "median_prefill_tps": 2061.856,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 1,
          "request_count_recorded_cases": 1,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": 1,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 46.76,
              "min": 46.76,
              "median": 46.76,
              "mean": 46.76,
              "p95": 46.76,
              "max": 46.76
            },
            "prefill_tps": {
              "n": 1,
              "total": 2061.855670103093,
              "min": 2061.855670103093,
              "median": 2061.855670103093,
              "mean": 2061.855670103093,
              "p95": 2061.855670103093,
              "max": 2061.855670103093
            },
            "elapsed_s": {
              "n": 1,
              "total": 3.901,
              "min": 3.901,
              "median": 3.901,
              "mean": 3.901,
              "p95": 3.901,
              "max": 3.901
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 621,
              "min": 621,
              "median": 621,
              "mean": 621,
              "p95": 621,
              "max": 621
            },
            "completion_tokens": {
              "n": 1,
              "total": 169,
              "min": 169,
              "median": 169,
              "mean": 169,
              "p95": 169,
              "max": 169
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-5",
      "run": "20260930T000829Z-smoke/flash.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 1,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 1,
      "strict": null,
      "prefill": null,
      "decode": 27.9,
      "ttft": null,
      "latency": 77.3,
      "wall": 77.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 65536,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T000829Z",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured server tensor split: 34,14",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": null,
        "checks_earned": 5,
        "checks_possible": 5,
        "check_percent": 100,
        "wall_s": 77.3,
        "median_decode_tps": 27.9,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 27.9,
            "min": 27.9,
            "median": 27.9,
            "mean": 27.9,
            "p95": 27.9,
            "max": 27.9
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 1,
            "total": 77.3,
            "min": 77.3,
            "median": 77.3,
            "mean": 77.3,
            "p95": 77.3,
            "max": 77.3
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 1,
            "total": 1594,
            "min": 1594,
            "median": 1594,
            "mean": 1594,
            "p95": 1594,
            "max": 1594
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 77.3,
          "median_decode_tps": 27.9,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 27.9,
              "min": 27.9,
              "median": 27.9,
              "mean": 27.9,
              "p95": 27.9,
              "max": 27.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 77.3,
              "min": 77.3,
              "median": 77.3,
              "mean": 77.3,
              "p95": 77.3,
              "max": 77.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1594,
              "min": 1594,
              "median": 1594,
              "mean": 1594,
              "p95": 1594,
              "max": 1594
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-6",
      "run": "20260930T001207Z-all/27b.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 91.837,
      "structure": 118,
      "strict": 82,
      "prefill": 611.0971703544069,
      "decode": 47.31,
      "ttft": null,
      "latency": 5.017,
      "wall": 1810.926,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": "q4_0",
      "date": "20260930T001207Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 118,
        "strict_passes": 82,
        "checks_earned": 675,
        "checks_possible": 735,
        "check_percent": 91.837,
        "wall_s": 1810.926,
        "median_decode_tps": 47.31,
        "median_prefill_tps": 611.097,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 5618.91,
            "min": 24.96,
            "median": 47.31,
            "mean": 45.68219512195122,
            "p95": 52.22,
            "max": 53.42
          },
          "prefill_tps": {
            "n": 123,
            "total": 76171.1886999845,
            "min": 290.76382392354174,
            "median": 611.0971703544069,
            "mean": 619.2779569104431,
            "p95": 798.2333555737774,
            "max": 805.5348298156633
          },
          "elapsed_s": {
            "n": 123,
            "total": 1810.926,
            "min": 0.996,
            "median": 5.017,
            "mean": 14.722975609756098,
            "p95": 65.023,
            "max": 102.419
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 63847,
            "min": 16,
            "median": 148,
            "mean": 519.0813008130082,
            "p95": 2169,
            "max": 4105
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 51,
          "strict_passes": 40,
          "checks_earned": 197,
          "checks_possible": 213,
          "check_percent": 92.488,
          "wall_s": 164.302,
          "median_decode_tps": 44.915,
          "median_prefill_tps": 533.403,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 2368.44,
              "min": 34.74,
              "median": 44.915,
              "mean": 45.54692307692308,
              "p95": 52.77,
              "max": 53.42
            },
            "prefill_tps": {
              "n": 52,
              "total": 26772.634815444566,
              "min": 290.76382392354174,
              "median": 533.4026426297743,
              "mean": 514.8583618354724,
              "p95": 666.2723201227241,
              "max": 743.4760700900448
            },
            "elapsed_s": {
              "n": 52,
              "total": 164.302,
              "min": 1.091,
              "median": 2.5955,
              "mean": 3.159653846153846,
              "p95": 7.299,
              "max": 7.952
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 6137,
              "min": 30,
              "median": 80.5,
              "mean": 118.01923076923077,
              "p95": 342,
              "max": 378
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 2,
          "checks_earned": 26,
          "checks_possible": 30,
          "check_percent": 86.667,
          "wall_s": 196.614,
          "median_decode_tps": 29.82,
          "median_prefill_tps": 656.66,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 85.3,
              "min": 24.96,
              "median": 29.82,
              "mean": 28.433333333333334,
              "p95": 30.52,
              "max": 30.52
            },
            "prefill_tps": {
              "n": 3,
              "total": 1960.4547271023241,
              "min": 599.0401568535572,
              "median": 656.6597341928199,
              "mean": 653.484909034108,
              "p95": 704.7548360559471,
              "max": 704.7548360559471
            },
            "elapsed_s": {
              "n": 3,
              "total": 196.614,
              "min": 42.209,
              "median": 60.349,
              "mean": 65.538,
              "p95": 94.056,
              "max": 94.056
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 570,
              "min": 147,
              "median": 167,
              "mean": 190,
              "p95": 256,
              "max": 256
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 37,
          "strict_passes": 18,
          "checks_earned": 211,
          "checks_possible": 236,
          "check_percent": 89.407,
          "wall_s": 1294.09,
          "median_decode_tps": 49.22,
          "median_prefill_tps": 785.247,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1860.16,
              "min": 36.47,
              "median": 49.22,
              "mean": 47.69641025641025,
              "p95": 51.06,
              "max": 51.06
            },
            "prefill_tps": {
              "n": 39,
              "total": 30518.252174272435,
              "min": 689.3324302637388,
              "median": 785.2472768488932,
              "mean": 782.519286519806,
              "p95": 804.8977598001328,
              "max": 805.5348298156633
            },
            "elapsed_s": {
              "n": 39,
              "total": 1294.09,
              "min": 5.267,
              "median": 26.549,
              "mean": 33.18179487179487,
              "p95": 86.513,
              "max": 102.419
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 51828,
              "min": 54,
              "median": 1122,
              "mean": 1328.923076923077,
              "p95": 3389,
              "max": 4105
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 27,
          "strict_passes": 22,
          "checks_earned": 241,
          "checks_possible": 256,
          "check_percent": 94.141,
          "wall_s": 155.92000000000002,
          "median_decode_tps": 45.61,
          "median_prefill_tps": 592.667,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1305.01,
              "min": 35.65,
              "median": 45.61,
              "mean": 45.000344827586204,
              "p95": 52.22,
              "max": 52.7
            },
            "prefill_tps": {
              "n": 29,
              "total": 16919.84698316518,
              "min": 367.61380710778377,
              "median": 592.666583426978,
              "mean": 583.442999419489,
              "p95": 783.0039396099044,
              "max": 789.1281391301274
            },
            "elapsed_s": {
              "n": 29,
              "total": 155.92000000000002,
              "min": 0.996,
              "median": 3.457,
              "mean": 5.376551724137931,
              "p95": 15.996,
              "max": 20.181
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 5312,
              "min": 16,
              "median": 129,
              "mean": 183.17241379310346,
              "p95": 699,
              "max": 952
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-7",
      "run": "20260930T001207Z-all/9b.jsonl",
      "model": "Qwen3.5-9B · Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 84.626,
      "structure": 106,
      "strict": 60,
      "prefill": 2054.719661918467,
      "decode": 46.62,
      "ttft": null,
      "latency": 4.065,
      "wall": 2147.694,
      "requests": 125,
      "retries": 0,
      "truncated": 2,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T001207Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured Docker GPU device allocation for this run.",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 106,
        "strict_passes": 60,
        "checks_earned": 622,
        "checks_possible": 735,
        "check_percent": 84.626,
        "wall_s": 2147.694,
        "median_decode_tps": 46.62,
        "median_prefill_tps": 2054.72,
        "retries": 0,
        "truncated": 2,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 5699.7,
            "min": 39.25,
            "median": 46.62,
            "mean": 46.3390243902439,
            "p95": 46.88,
            "max": 47.02
          },
          "prefill_tps": {
            "n": 123,
            "total": 253483.24586918944,
            "min": 947.7305972763047,
            "median": 2054.719661918467,
            "mean": 2060.8393973104835,
            "p95": 2659.0466422071845,
            "max": 2688.902395643978
          },
          "elapsed_s": {
            "n": 123,
            "total": 2147.694,
            "min": 0.706,
            "median": 4.065,
            "mean": 17.460926829268292,
            "p95": 74.327,
            "max": 163.413
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 90784,
            "min": 8,
            "median": 172,
            "mean": 738.0813008130082,
            "p95": 3048,
            "max": 6000
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 33,
          "checks_earned": 185,
          "checks_possible": 213,
          "check_percent": 86.854,
          "wall_s": 173.624,
          "median_decode_tps": 46.68,
          "median_prefill_tps": 1774.087,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 2427.39,
              "min": 46.46,
              "median": 46.68,
              "mean": 46.68057692307692,
              "p95": 46.88,
              "max": 46.94
            },
            "prefill_tps": {
              "n": 52,
              "total": 88723.661919864,
              "min": 947.7305972763047,
              "median": 1774.0873803419418,
              "mean": 1706.2242676896924,
              "p95": 2185.2448007163334,
              "max": 2419.6826056957057
            },
            "elapsed_s": {
              "n": 52,
              "total": 173.624,
              "min": 0.953,
              "median": 2.158,
              "mean": 3.3389230769230767,
              "p95": 8.998,
              "max": 12.604
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7674,
              "min": 35,
              "median": 91.5,
              "mean": 147.57692307692307,
              "p95": 414,
              "max": 582
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 1,
          "strict_passes": 0,
          "checks_earned": 18,
          "checks_possible": 30,
          "check_percent": 60,
          "wall_s": 384.548,
          "median_decode_tps": 40.87,
          "median_prefill_tps": 2168.348,
          "retries": 0,
          "truncated": 2,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 122.36,
              "min": 39.25,
              "median": 40.87,
              "mean": 40.78666666666667,
              "p95": 42.24,
              "max": 42.24
            },
            "prefill_tps": {
              "n": 3,
              "total": 6469.073679713452,
              "min": 1969.1439556562655,
              "median": 2168.347982303617,
              "mean": 2156.3578932378173,
              "p95": 2331.58174175357,
              "max": 2331.58174175357
            },
            "elapsed_s": {
              "n": 3,
              "total": 384.548,
              "min": 67.758,
              "median": 153.377,
              "mean": 128.18266666666668,
              "p95": 163.413,
              "max": 163.413
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 13654,
              "min": 1654,
              "median": 6000,
              "mean": 4551.333333333333,
              "p95": 6000,
              "max": 6000
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 28,
          "strict_passes": 12,
          "checks_earned": 190,
          "checks_possible": 236,
          "check_percent": 80.508,
          "wall_s": 1453.113,
          "median_decode_tps": 46.42,
          "median_prefill_tps": 2623.675,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1798.5,
              "min": 41.95,
              "median": 46.42,
              "mean": 46.11538461538461,
              "p95": 46.74,
              "max": 46.78
            },
            "prefill_tps": {
              "n": 39,
              "total": 101884.49020897641,
              "min": 2283.3699155821373,
              "median": 2623.6754602369915,
              "mean": 2612.4228258711896,
              "p95": 2684.6790192208373,
              "max": 2688.902395643978
            },
            "elapsed_s": {
              "n": 39,
              "total": 1453.113,
              "min": 2.689,
              "median": 30.326,
              "mean": 37.259307692307694,
              "p95": 98.784,
              "max": 99.101
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 63690,
              "min": 64,
              "median": 1378,
              "mean": 1633.076923076923,
              "p95": 4098,
              "max": 4455
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 25,
          "strict_passes": 15,
          "checks_earned": 229,
          "checks_possible": 256,
          "check_percent": 89.453,
          "wall_s": 136.409,
          "median_decode_tps": 46.67,
          "median_prefill_tps": 1974.268,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1351.45,
              "min": 45.21,
              "median": 46.67,
              "mean": 46.601724137931036,
              "p95": 46.98,
              "max": 47.02
            },
            "prefill_tps": {
              "n": 29,
              "total": 56406.0200606356,
              "min": 1202.3015486788995,
              "median": 1974.267968056788,
              "mean": 1945.035174504676,
              "p95": 2616.402900098818,
              "max": 2631.330811027427
            },
            "elapsed_s": {
              "n": 29,
              "total": 136.409,
              "min": 0.706,
              "median": 3.246,
              "mean": 4.703758620689655,
              "p95": 18.993,
              "max": 23.01
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 5766,
              "min": 8,
              "median": 131,
              "mean": 198.82758620689654,
              "p95": 859,
              "max": 1047
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-8",
      "run": "20260930T001207Z-all/flash.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 89.231,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 28.3,
      "ttft": null,
      "latency": 58.5,
      "wall": 1575.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 65536,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T001207Z",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured server tensor split: 34,14",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 116,
        "checks_possible": 130,
        "check_percent": 89.231,
        "wall_s": 1575.6,
        "median_decode_tps": 28.3,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 8,
        "fabricated_quotes": 6,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 678.3,
            "min": 27.3,
            "median": 28.3,
            "mean": 28.2625,
            "p95": 28.9,
            "max": 28.9
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1575.6,
            "min": 33.5,
            "median": 58.5,
            "mean": 65.65,
            "p95": 101.5,
            "max": 163.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 33764,
            "min": 744,
            "median": 1218.5,
            "mean": 1406.8333333333333,
            "p95": 2085,
            "max": 3931
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 63,
          "check_percent": 98.413,
          "wall_s": 645,
          "median_decode_tps": 28.55,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 342.2,
              "min": 27.8,
              "median": 28.55,
              "mean": 28.516666666666666,
              "p95": 28.9,
              "max": 28.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 645,
              "min": 33.5,
              "median": 45.45,
              "mean": 53.75,
              "p95": 94.9,
              "max": 94.9
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13801,
              "min": 744,
              "median": 965,
              "mean": 1150.0833333333333,
              "p95": 1930,
              "max": 1930
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 54,
          "checks_possible": 67,
          "check_percent": 80.597,
          "wall_s": 930.6,
          "median_decode_tps": 28.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 8,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 336.1,
              "min": 27.3,
              "median": 28.1,
              "mean": 28.008333333333333,
              "p95": 28.6,
              "max": 28.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 930.6,
              "min": 47.6,
              "median": 67.6,
              "mean": 77.55,
              "p95": 163.5,
              "max": 163.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 19963,
              "min": 1026,
              "median": 1516,
              "mean": 1663.5833333333333,
              "p95": 3931,
              "max": 3931
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-9",
      "run": "20260930T171003Z-35b-single-gpu-cpu40/35b.jsonl",
      "model": "Qwen3.6-35B-A3B · IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 86.259,
      "structure": 104,
      "strict": 65,
      "prefill": 325.34412368399677,
      "decode": 36.11,
      "ttft": null,
      "latency": 7.235,
      "wall": 3069.038,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": null,
      "date": "20260930T171003Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Captured server command: --split-mode none (single model GPU).",
      "notes": "",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 104,
        "strict_passes": 65,
        "checks_earned": 634,
        "checks_possible": 735,
        "check_percent": 86.259,
        "wall_s": 3069.038,
        "median_decode_tps": 36.11,
        "median_prefill_tps": 325.344,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 4459.62,
            "min": 30.9,
            "median": 36.11,
            "mean": 36.25707317073171,
            "p95": 41.91,
            "max": 42.05
          },
          "prefill_tps": {
            "n": 123,
            "total": 37031.1354370793,
            "min": 100.92544242642316,
            "median": 325.34412368399677,
            "mean": 301.0661417648723,
            "p95": 429.40459007461163,
            "max": 439.60387066302866
          },
          "elapsed_s": {
            "n": 123,
            "total": 3069.038,
            "min": 1.555,
            "median": 7.235,
            "mean": 24.951528455284553,
            "p95": 98.997,
            "max": 144.637
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 79608,
            "min": 16,
            "median": 197,
            "mean": 647.219512195122,
            "p95": 2424,
            "max": 4313
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 51,
          "strict_passes": 36,
          "checks_earned": 195,
          "checks_possible": 213,
          "check_percent": 91.549,
          "wall_s": 278.618,
          "median_decode_tps": 34.805,
          "median_prefill_tps": 244.603,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 1805.23,
              "min": 31.46,
              "median": 34.805,
              "mean": 34.71596153846154,
              "p95": 37.91,
              "max": 40.57
            },
            "prefill_tps": {
              "n": 52,
              "total": 12375.551016861571,
              "min": 100.92544242642316,
              "median": 244.60332812176986,
              "mean": 237.99136570887634,
              "p95": 374.0509411579701,
              "max": 413.384285099515
            },
            "elapsed_s": {
              "n": 52,
              "total": 278.618,
              "min": 2.254,
              "median": 3.9985,
              "mean": 5.358038461538461,
              "p95": 10.778,
              "max": 20.803
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7390,
              "min": 45,
              "median": 94,
              "mean": 142.1153846153846,
              "p95": 379,
              "max": 624
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 1,
          "checks_earned": 28,
          "checks_possible": 30,
          "check_percent": 93.333,
          "wall_s": 316.93399999999997,
          "median_decode_tps": 37.13,
          "median_prefill_tps": 421.91,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 106.49000000000001,
              "min": 30.9,
              "median": 37.13,
              "mean": 35.49666666666667,
              "p95": 38.46,
              "max": 38.46
            },
            "prefill_tps": {
              "n": 3,
              "total": 1230.647049917962,
              "min": 379.4155000937413,
              "median": 421.91037907161154,
              "mean": 410.21568330598734,
              "p95": 429.32117075260913,
              "max": 429.32117075260913
            },
            "elapsed_s": {
              "n": 3,
              "total": 316.93399999999997,
              "min": 77.682,
              "median": 94.615,
              "mean": 105.64466666666667,
              "p95": 144.637,
              "max": 144.637
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 1382,
              "min": 349,
              "median": 400,
              "mean": 460.6666666666667,
              "p95": 633,
              "max": 633
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 23,
          "strict_passes": 14,
          "checks_earned": 185,
          "checks_possible": 236,
          "check_percent": 78.39,
          "wall_s": 2180.034,
          "median_decode_tps": 34.8,
          "median_prefill_tps": 408.809,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1359.59,
              "min": 31.29,
              "median": 34.8,
              "mean": 34.86128205128205,
              "p95": 37.18,
              "max": 37.33
            },
            "prefill_tps": {
              "n": 39,
              "total": 15807.25683874502,
              "min": 356.4133517370042,
              "median": 408.80880306692427,
              "mean": 405.31427791653897,
              "p95": 433.89076473984096,
              "max": 439.60387066302866
            },
            "elapsed_s": {
              "n": 39,
              "total": 2180.034,
              "min": 9.205,
              "median": 45.648,
              "mean": 55.89830769230769,
              "p95": 137.399,
              "max": 141.903
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 62218,
              "min": 46,
              "median": 1473,
              "mean": 1595.3333333333333,
              "p95": 3522,
              "max": 4313
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 27,
          "strict_passes": 14,
          "checks_earned": 226,
          "checks_possible": 256,
          "check_percent": 88.281,
          "wall_s": 293.452,
          "median_decode_tps": 41.69,
          "median_prefill_tps": 268.124,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1188.31,
              "min": 35.34,
              "median": 41.69,
              "mean": 40.97620689655172,
              "p95": 42.02,
              "max": 42.05
            },
            "prefill_tps": {
              "n": 29,
              "total": 7617.680531554744,
              "min": 128.5817938384829,
              "median": 268.12441678529404,
              "mean": 262.67863901912915,
              "p95": 428.7822254057367,
              "max": 432.7522669711866
            },
            "elapsed_s": {
              "n": 29,
              "total": 293.452,
              "min": 1.555,
              "median": 6.091,
              "mean": 10.11903448275862,
              "p95": 38.977,
              "max": 58.68
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 8618,
              "min": 16,
              "median": 143,
              "mean": 297.17241379310343,
              "p95": 1436,
              "max": 1473
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-10",
      "run": "20260930T193322Z-lab system-flash-next-lmstudio/flash.jsonl",
      "model": "Qwen3.8-Flash-Next-UD-Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "UD-Q2_K_XL",
      "runtime": "llama.cpp-win-x86_64-nvidia-cuda12-avx2@2.42.0",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 88.707,
      "structure": 117,
      "strict": 80,
      "prefill": 271.95,
      "decode": 18.07,
      "ttft": null,
      "latency": 12.181,
      "wall": 5468.709,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 1,
      "kv": null,
      "date": "20260930T193322Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 5090 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core Ultra · RTX 5090",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 117,
        "strict_passes": 80,
        "checks_earned": 652,
        "checks_possible": 735,
        "check_percent": 88.707,
        "wall_s": 5468.709,
        "median_decode_tps": 18.07,
        "median_prefill_tps": 271.95,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 2169.91,
            "min": 12.61,
            "median": 18.07,
            "mean": 17.641544715447154,
            "p95": 18.61,
            "max": 18.72
          },
          "prefill_tps": {
            "n": 123,
            "total": 31795.52,
            "min": 83.62,
            "median": 271.95,
            "mean": 258.500162601626,
            "p95": 383.88,
            "max": 398.82
          },
          "elapsed_s": {
            "n": 123,
            "total": 5468.709,
            "min": 2.287,
            "median": 12.181,
            "mean": 44.46104878048781,
            "p95": 190.408,
            "max": 360.991
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 75177,
            "min": 14,
            "median": 166,
            "mean": 611.1951219512196,
            "p95": 2798,
            "max": 4232
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 42,
          "checks_earned": 201,
          "checks_possible": 213,
          "check_percent": 94.366,
          "wall_s": 436.052,
          "median_decode_tps": 18.27,
          "median_prefill_tps": 179.225,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 949.27,
              "min": 17.6,
              "median": 18.27,
              "mean": 18.255192307692308,
              "p95": 18.64,
              "max": 18.67
            },
            "prefill_tps": {
              "n": 52,
              "total": 10431.45,
              "min": 83.62,
              "median": 179.225,
              "mean": 200.6048076923077,
              "p95": 325.56,
              "max": 351.3
            },
            "elapsed_s": {
              "n": 52,
              "total": 436.052,
              "min": 2.287,
              "median": 5.4675,
              "mean": 8.385615384615384,
              "p95": 24.398,
              "max": 43.395
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 6509,
              "min": 14,
              "median": 73,
              "mean": 125.17307692307692,
              "p95": 417,
              "max": 750
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 2,
          "checks_earned": 29,
          "checks_possible": 30,
          "check_percent": 96.667,
          "wall_s": 558.173,
          "median_decode_tps": 13.84,
          "median_prefill_tps": 371.57,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 40.92,
              "min": 12.61,
              "median": 13.84,
              "mean": 13.64,
              "p95": 14.47,
              "max": 14.47
            },
            "prefill_tps": {
              "n": 3,
              "total": 1105.26,
              "min": 361.71,
              "median": 371.57,
              "mean": 368.42,
              "p95": 371.98,
              "max": 371.98
            },
            "elapsed_s": {
              "n": 3,
              "total": 558.173,
              "min": 83.734,
              "median": 113.448,
              "mean": 186.05766666666665,
              "p95": 360.991,
              "max": 360.991
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 3215,
              "min": 185,
              "median": 232,
              "mean": 1071.6666666666667,
              "p95": 2798,
              "max": 2798
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 36,
          "strict_passes": 16,
          "checks_earned": 188,
          "checks_possible": 236,
          "check_percent": 79.661,
          "wall_s": 4020.576,
          "median_decode_tps": 17,
          "median_prefill_tps": 351.87,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 657.69,
              "min": 14.1,
              "median": 17,
              "mean": 16.863846153846154,
              "p95": 17.57,
              "max": 17.59
            },
            "prefill_tps": {
              "n": 39,
              "total": 13646.92,
              "min": 292.43,
              "median": 351.87,
              "mean": 349.92102564102566,
              "p95": 390.36,
              "max": 398.82
            },
            "elapsed_s": {
              "n": 39,
              "total": 4020.576,
              "min": 12.181,
              "median": 81.623,
              "mean": 103.09169230769231,
              "p95": 304.863,
              "max": 316.796
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 59090,
              "min": 46,
              "median": 1293,
              "mean": 1515.128205128205,
              "p95": 3976,
              "max": 4232
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 26,
          "strict_passes": 20,
          "checks_earned": 234,
          "checks_possible": 256,
          "check_percent": 91.406,
          "wall_s": 453.908,
          "median_decode_tps": 18.16,
          "median_prefill_tps": 243.51,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 522.03,
              "min": 16.05,
              "median": 18.16,
              "mean": 18.00103448275862,
              "p95": 18.7,
              "max": 18.72
            },
            "prefill_tps": {
              "n": 29,
              "total": 6611.89,
              "min": 101.57,
              "median": 243.51,
              "mean": 227.9962068965517,
              "p95": 368.93,
              "max": 387.7
            },
            "elapsed_s": {
              "n": 29,
              "total": 453.908,
              "min": 2.499,
              "median": 9.548,
              "mean": 15.652,
              "p95": 57.411,
              "max": 109.381
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 6363,
              "min": 16,
              "median": 90,
              "mean": 219.41379310344828,
              "p95": 937,
              "max": 1878
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-11",
      "run": "20260930T232800Z-b70-27b-q4/27b-q4.jsonl",
      "model": "Qwen3.8-27B-UD-Q4_K_XL",
      "model_family": "27B",
      "quant": "UD-Q4_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 91.156,
      "structure": 119,
      "strict": 82,
      "prefill": 416.45927421012016,
      "decode": 33.8,
      "ttft": null,
      "latency": 8.925,
      "wall": 2597.417,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T232800Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 119,
        "strict_passes": 82,
        "checks_earned": 670,
        "checks_possible": 735,
        "check_percent": 91.156,
        "wall_s": 2597.417,
        "median_decode_tps": 33.8,
        "median_prefill_tps": 416.459,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 4050,
            "min": 23.6,
            "median": 33.8,
            "mean": 32.926829268292686,
            "p95": 37.22,
            "max": 37.87
          },
          "prefill_tps": {
            "n": 123,
            "total": 59594.801437197646,
            "min": 103.67540582723396,
            "median": 416.45927421012016,
            "mean": 484.51058079022476,
            "p95": 815.4851657264549,
            "max": 857.7154807628152
          },
          "elapsed_s": {
            "n": 123,
            "total": 2597.417,
            "min": 1.798,
            "median": 8.925,
            "mean": 21.117211382113823,
            "p95": 74.157,
            "max": 122.624
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 72012,
            "min": 10,
            "median": 207,
            "mean": 585.4634146341464,
            "p95": 2219,
            "max": 4155
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 42,
          "checks_earned": 201,
          "checks_possible": 213,
          "check_percent": 94.366,
          "wall_s": 276.718,
          "median_decode_tps": 32.195,
          "median_prefill_tps": 300.195,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 1695.19,
              "min": 26.94,
              "median": 32.195,
              "mean": 32.59980769230769,
              "p95": 37.43,
              "max": 37.87
            },
            "prefill_tps": {
              "n": 52,
              "total": 15739.427041639046,
              "min": 103.67540582723396,
              "median": 300.1951978800836,
              "mean": 302.68128926228934,
              "p95": 582.8962296726245,
              "max": 693.9305473106125
            },
            "elapsed_s": {
              "n": 52,
              "total": 276.718,
              "min": 2.11,
              "median": 4.1754999999999995,
              "mean": 5.3215,
              "p95": 11.705,
              "max": 12.732
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7202,
              "min": 36,
              "median": 98,
              "mean": 138.5,
              "p95": 376,
              "max": 424
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 2,
          "strict_passes": 2,
          "checks_earned": 25,
          "checks_possible": 30,
          "check_percent": 83.333,
          "wall_s": 162.593,
          "median_decode_tps": 24,
          "median_prefill_tps": 792.361,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 74.89,
              "min": 23.6,
              "median": 24,
              "mean": 24.963333333333335,
              "p95": 27.29,
              "max": 27.29
            },
            "prefill_tps": {
              "n": 3,
              "total": 2370.518822693412,
              "min": 768.8197723948766,
              "median": 792.3611701633138,
              "mean": 790.172940897804,
              "p95": 809.3378801352216,
              "max": 809.3378801352216
            },
            "elapsed_s": {
              "n": 3,
              "total": 162.593,
              "min": 41.56,
              "median": 54.549,
              "mean": 54.19766666666666,
              "p95": 66.484,
              "max": 66.484
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 439,
              "min": 10,
              "median": 213,
              "mean": 146.33333333333334,
              "p95": 216,
              "max": 216
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 36,
          "strict_passes": 16,
          "checks_earned": 204,
          "checks_possible": 236,
          "check_percent": 86.441,
          "wall_s": 1897.4759999999999,
          "median_decode_tps": 34.62,
          "median_prefill_tps": 782.045,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1325.13,
              "min": 25.86,
              "median": 34.62,
              "mean": 33.97769230769231,
              "p95": 36.25,
              "max": 36.47
            },
            "prefill_tps": {
              "n": 39,
              "total": 30152.85000133297,
              "min": 648.1681914625225,
              "median": 782.0445822435963,
              "mean": 773.1500000341787,
              "p95": 845.5690846609101,
              "max": 857.7154807628152
            },
            "elapsed_s": {
              "n": 39,
              "total": 1897.4759999999999,
              "min": 7.288,
              "median": 41.459,
              "mean": 48.65323076923077,
              "p95": 118.01,
              "max": 122.624
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 57533,
              "min": 58,
              "median": 1284,
              "mean": 1475.2051282051282,
              "p95": 3268,
              "max": 4155
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 29,
          "strict_passes": 22,
          "checks_earned": 240,
          "checks_possible": 256,
          "check_percent": 93.75,
          "wall_s": 260.63,
          "median_decode_tps": 32.48,
          "median_prefill_tps": 389.119,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 954.79,
              "min": 26.9,
              "median": 32.48,
              "mean": 32.923793103448276,
              "p95": 37.26,
              "max": 37.34
            },
            "prefill_tps": {
              "n": 29,
              "total": 11332.00557153222,
              "min": 140.63739994932587,
              "median": 389.11898938355256,
              "mean": 390.7588128114558,
              "p95": 727.6994693406982,
              "max": 807.0063563971089
            },
            "elapsed_s": {
              "n": 29,
              "total": 260.63,
              "min": 1.798,
              "median": 5.675,
              "mean": 8.987241379310344,
              "p95": 29.679,
              "max": 50.332
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 6838,
              "min": 10,
              "median": 138,
              "mean": 235.79310344827587,
              "p95": 974,
              "max": 1705
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-12",
      "run": "knobs/9b-4060-cuda-rp1.1.jsonl",
      "model": "Qwen3.5-9B · Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 78.102,
      "structure": 20,
      "strict": null,
      "prefill": 2588.307396048218,
      "decode": 41.18693788217118,
      "ttft": null,
      "latency": 37.15,
      "wall": 922,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The full 24 items on the B70 / same 9B on one RTX 4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 20,
        "strict_passes": null,
        "checks_earned": 107,
        "checks_possible": 137,
        "check_percent": 78.102,
        "wall_s": 922,
        "median_decode_tps": 41.187,
        "median_prefill_tps": 2588.307,
        "retries": null,
        "truncated": 0,
        "near_quotes": 18,
        "fabricated_quotes": 12,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 985.927728184172,
            "min": 40.2048222846343,
            "median": 41.18693788217118,
            "mean": 41.080322007673836,
            "p95": 41.510441089444875,
            "max": 41.5292398258168
          },
          "prefill_tps": {
            "n": 24,
            "total": 62090.23078235421,
            "min": 2454.0544240420836,
            "median": 2588.307396048218,
            "mean": 2587.0929492647588,
            "p95": 2675.008183270577,
            "max": 2679.180791119785
          },
          "elapsed_s": {
            "n": 24,
            "total": 922,
            "min": 19.5,
            "median": 37.15,
            "mean": 38.416666666666664,
            "p95": 54.5,
            "max": 78.4
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 36788,
            "min": 766,
            "median": 1487.5,
            "mean": 1532.8333333333333,
            "p95": 2204,
            "max": 3149
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 54,
          "checks_possible": 63,
          "check_percent": 85.714,
          "wall_s": 386.6,
          "median_decode_tps": 41.148,
          "median_prefill_tps": 2560.699,
          "retries": null,
          "truncated": 0,
          "near_quotes": 8,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 492.2109081690844,
              "min": 40.2048222846343,
              "median": 41.148463124969965,
              "mean": 41.017575680757034,
              "p95": 41.5292398258168,
              "max": 41.5292398258168
            },
            "prefill_tps": {
              "n": 12,
              "total": 30772.088203695526,
              "min": 2454.0544240420836,
              "median": 2560.698817743578,
              "mean": 2564.3406836412937,
              "p95": 2679.180791119785,
              "max": 2679.180791119785
            },
            "elapsed_s": {
              "n": 12,
              "total": 386.6,
              "min": 19.5,
              "median": 28.95,
              "mean": 32.21666666666667,
              "p95": 53.2,
              "max": 53.2
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 15381,
              "min": 766,
              "median": 1145,
              "mean": 1281.75,
              "p95": 2107,
              "max": 2107
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 9,
          "strict_passes": null,
          "checks_earned": 53,
          "checks_possible": 74,
          "check_percent": 71.622,
          "wall_s": 535.4,
          "median_decode_tps": 41.235,
          "median_prefill_tps": 2618.107,
          "retries": null,
          "truncated": 0,
          "near_quotes": 10,
          "fabricated_quotes": 11,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 493.7168200150876,
              "min": 40.76917438766377,
              "median": 41.23457953085779,
              "mean": 41.14306833459063,
              "p95": 41.3913502625554,
              "max": 41.3913502625554
            },
            "prefill_tps": {
              "n": 12,
              "total": 31318.14257865868,
              "min": 2536.908982511374,
              "median": 2618.106525176668,
              "mean": 2609.8452148882234,
              "p95": 2675.008183270577,
              "max": 2675.008183270577
            },
            "elapsed_s": {
              "n": 12,
              "total": 535.4,
              "min": 29.3,
              "median": 42.4,
              "mean": 44.61666666666667,
              "p95": 78.4,
              "max": 78.4
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 21407,
              "min": 1160,
              "median": 1668.5,
              "mean": 1783.9166666666667,
              "p95": 3149,
              "max": 3149
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-13",
      "run": "knobs/9b-a770-opencl-rp1.1.jsonl",
      "model": "Qwen3.5-9B · Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp SYCL · OpenCL runtime",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 87.313,
      "structure": 20,
      "strict": null,
      "prefill": 1397.712327812339,
      "decode": 35.47663786086463,
      "ttft": null,
      "latency": 36.3,
      "wall": 1123.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 20,
        "strict_passes": null,
        "checks_earned": 117,
        "checks_possible": 134,
        "check_percent": 87.313,
        "wall_s": 1123.3,
        "median_decode_tps": 35.477,
        "median_prefill_tps": 1397.712,
        "retries": null,
        "truncated": 0,
        "near_quotes": 9,
        "fabricated_quotes": 8,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 838.553116778698,
            "min": 32.660664777925156,
            "median": 35.47663786086463,
            "mean": 34.93971319911242,
            "p95": 36.46013075659076,
            "max": 36.555689277342786
          },
          "prefill_tps": {
            "n": 24,
            "total": 31859.907283445234,
            "min": 36.515122757308426,
            "median": 1397.712327812339,
            "mean": 1327.496136810218,
            "p95": 1472.0695983896608,
            "max": 1488.9705771450256
          },
          "elapsed_s": {
            "n": 24,
            "total": 1123.3,
            "min": 23,
            "median": 36.3,
            "mean": 46.80416666666667,
            "p95": 101.5,
            "max": 144.8
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 34217,
            "min": 754,
            "median": 1250.5,
            "mean": 1425.7083333333333,
            "p95": 2520,
            "max": 3246
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 55,
          "checks_possible": 62,
          "check_percent": 88.71,
          "wall_s": 535.4,
          "median_decode_tps": 35.587,
          "median_prefill_tps": 1354.951,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 424.7607627903551,
              "min": 33.04253703686808,
              "median": 35.587461120144994,
              "mean": 35.39673023252959,
              "p95": 36.555689277342786,
              "max": 36.555689277342786
            },
            "prefill_tps": {
              "n": 12,
              "total": 14863.54544757949,
              "min": 36.515122757308426,
              "median": 1354.9510620659728,
              "mean": 1238.628787298291,
              "p95": 1472.0695983896608,
              "max": 1472.0695983896608
            },
            "elapsed_s": {
              "n": 12,
              "total": 535.4,
              "min": 23,
              "median": 34.8,
              "mean": 44.61666666666667,
              "p95": 144.8,
              "max": 144.8
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 14991,
              "min": 754,
              "median": 1185.5,
              "mean": 1249.25,
              "p95": 1906,
              "max": 1906
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 9,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 72,
          "check_percent": 86.111,
          "wall_s": 587.9,
          "median_decode_tps": 34.611,
          "median_prefill_tps": 1405.748,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 413.792353988343,
              "min": 32.660664777925156,
              "median": 34.61130499481972,
              "mean": 34.48269616569525,
              "p95": 35.80560181950227,
              "max": 35.80560181950227
            },
            "prefill_tps": {
              "n": 12,
              "total": 16996.361835865744,
              "min": 1359.8360485566757,
              "median": 1405.7479984625154,
              "mean": 1416.3634863221455,
              "p95": 1488.9705771450256,
              "max": 1488.9705771450256
            },
            "elapsed_s": {
              "n": 12,
              "total": 587.9,
              "min": 29.8,
              "median": 41.7,
              "mean": 48.99166666666667,
              "p95": 101.5,
              "max": 101.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 19226,
              "min": 1009,
              "median": 1377.5,
              "mean": 1602.1666666666667,
              "p95": 3246,
              "max": 3246
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-14",
      "run": "knobs/9b-fast-lab-rp1.1.jsonl",
      "model": "Qwen3.5-9B Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "as above, request `repeat_penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 84.444,
      "structure": 20,
      "strict": null,
      "prefill": 41.45121351875794,
      "decode": 25.03872689903599,
      "ttft": null,
      "latency": 49.15,
      "wall": 1383.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 61368,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 20,
        "strict_passes": null,
        "checks_earned": 114,
        "checks_possible": 135,
        "check_percent": 84.444,
        "wall_s": 1383.8,
        "median_decode_tps": 25.039,
        "median_prefill_tps": 41.451,
        "retries": null,
        "truncated": 0,
        "near_quotes": 12,
        "fabricated_quotes": 9,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 589.891655326458,
            "min": 22.019465019105848,
            "median": 25.03872689903599,
            "mean": 24.57881897193575,
            "p95": 25.776912482205827,
            "max": 25.864368993814978
          },
          "prefill_tps": {
            "n": 24,
            "total": 974.5143891724449,
            "min": 37.399605434162666,
            "median": 41.45121351875794,
            "mean": 40.60476621551854,
            "p95": 41.96787359276473,
            "max": 41.98373130411965
          },
          "elapsed_s": {
            "n": 24,
            "total": 1383.8,
            "min": 29.9,
            "median": 49.15,
            "mean": 57.65833333333333,
            "p95": 101.2,
            "max": 104
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 33569,
            "min": 754,
            "median": 1250.5,
            "mean": 1398.7083333333333,
            "p95": 2375,
            "max": 2520
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 61368,
            "min": 1215,
            "median": 2187,
            "mean": 2557,
            "p95": 4333,
            "max": 4932
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 55,
          "checks_possible": 62,
          "check_percent": 88.71,
          "wall_s": 615.2,
          "median_decode_tps": 25.438,
          "median_prefill_tps": 41.673,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 298.6625612648702,
              "min": 22.799108217914487,
              "median": 25.43825958095129,
              "mean": 24.888546772072512,
              "p95": 25.864368993814978,
              "max": 25.864368993814978
            },
            "prefill_tps": {
              "n": 12,
              "total": 493.4624812711,
              "min": 37.399605434162666,
              "median": 41.67274416345681,
              "mean": 41.12187343925833,
              "p95": 41.98373130411965,
              "max": 41.98373130411965
            },
            "elapsed_s": {
              "n": 12,
              "total": 615.2,
              "min": 29.9,
              "median": 46.7,
              "mean": 51.266666666666666,
              "p95": 85.9,
              "max": 85.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 15112,
              "min": 754,
              "median": 1185.5,
              "mean": 1259.3333333333333,
              "p95": 1999,
              "max": 1999
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 25984,
              "min": 1215,
              "median": 1902.5,
              "mean": 2165.3333333333335,
              "p95": 4333,
              "max": 4333
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 9,
          "strict_passes": null,
          "checks_earned": 59,
          "checks_possible": 73,
          "check_percent": 80.822,
          "wall_s": 768.6,
          "median_decode_tps": 24.621,
          "median_prefill_tps": 40.321,
          "retries": null,
          "truncated": 0,
          "near_quotes": 8,
          "fabricated_quotes": 6,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 291.22909406158783,
              "min": 22.019465019105848,
              "median": 24.62051882036792,
              "mean": 24.269091171798983,
              "p95": 25.60393772210071,
              "max": 25.60393772210071
            },
            "prefill_tps": {
              "n": 12,
              "total": 481.0519079013449,
              "min": 37.50937734433608,
              "median": 40.32055894833533,
              "mean": 40.087658991778746,
              "p95": 41.876485306588215,
              "max": 41.876485306588215
            },
            "elapsed_s": {
              "n": 12,
              "total": 768.6,
              "min": 40.1,
              "median": 55.849999999999994,
              "mean": 64.05,
              "p95": 104,
              "max": 104
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 18457,
              "min": 961,
              "median": 1382,
              "mean": 1538.0833333333333,
              "p95": 2520,
              "max": 2520
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 35384,
              "min": 1830,
              "median": 2861.5,
              "mean": 2948.6666666666665,
              "p95": 4932,
              "max": 4932
            }
          }
        }
      }
    },
    {
      "id": "run-15",
      "run": "knobs/b70-moe-16gbsim-ncpu18-cap13g-cram0-all24-p1.jsonl",
      "model": "Flash-Next MoE Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "as above with a 13 GiB cap and `--cache-ram 0`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 92.913,
      "structure": 22,
      "strict": null,
      "prefill": 65.66446522104684,
      "decode": 16.390798219752256,
      "ttft": null,
      "latency": 110.4,
      "wall": 2934.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 118,
        "checks_possible": 127,
        "check_percent": 92.913,
        "wall_s": 2934.2,
        "median_decode_tps": 16.391,
        "median_prefill_tps": 65.664,
        "retries": null,
        "truncated": 0,
        "near_quotes": 9,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 394.98288952813704,
            "min": 15.67680566313098,
            "median": 16.390798219752256,
            "mean": 16.45762039700571,
            "p95": 17.2166594576103,
            "max": 17.498340811874613
          },
          "prefill_tps": {
            "n": 24,
            "total": 1550.4581924341467,
            "min": 45.880892155258834,
            "median": 65.66446522104684,
            "mean": 64.60242468475612,
            "p95": 69.57816775190018,
            "max": 74.4381897837454
          },
          "elapsed_s": {
            "n": 24,
            "total": 2934.2,
            "min": 64,
            "median": 110.4,
            "mean": 122.25833333333333,
            "p95": 206.6,
            "max": 209.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 32308,
            "min": 770,
            "median": 1247.5,
            "mean": 1346.1666666666667,
            "p95": 2392,
            "max": 2396
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 65,
          "check_percent": 95.385,
          "wall_s": 1263.5,
          "median_decode_tps": 16.493,
          "median_prefill_tps": 62.968,
          "retries": null,
          "truncated": 0,
          "near_quotes": 3,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 199.2643346901041,
              "min": 16.09039620911718,
              "median": 16.49278646696051,
              "mean": 16.605361224175343,
              "p95": 17.498340811874613,
              "max": 17.498340811874613
            },
            "prefill_tps": {
              "n": 12,
              "total": 742.5365889566365,
              "min": 45.880892155258834,
              "median": 62.96800139931062,
              "mean": 61.8780490797197,
              "p95": 66.8526810204642,
              "max": 66.8526810204642
            },
            "elapsed_s": {
              "n": 12,
              "total": 1263.5,
              "min": 64,
              "median": 88.55,
              "mean": 105.29166666666666,
              "p95": 200.1,
              "max": 200.1
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13718,
              "min": 770,
              "median": 959,
              "mean": 1143.1666666666667,
              "p95": 2082,
              "max": 2082
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 56,
          "checks_possible": 62,
          "check_percent": 90.323,
          "wall_s": 1670.7,
          "median_decode_tps": 16.384,
          "median_prefill_tps": 68.489,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 195.71855483803293,
              "min": 15.67680566313098,
              "median": 16.3835276255231,
              "mean": 16.309879569836077,
              "p95": 16.762139630673612,
              "max": 16.762139630673612
            },
            "prefill_tps": {
              "n": 12,
              "total": 807.9216034775103,
              "min": 53.30378762912523,
              "median": 68.48918780999446,
              "mean": 67.32680028979252,
              "p95": 74.4381897837454,
              "max": 74.4381897837454
            },
            "elapsed_s": {
              "n": 12,
              "total": 1670.7,
              "min": 100,
              "median": 133.75,
              "mean": 139.225,
              "p95": 209.7,
              "max": 209.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 18590,
              "min": 1083,
              "median": 1440.5,
              "mean": 1549.1666666666667,
              "p95": 2396,
              "max": 2396
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-16",
      "run": "knobs/b70-moe-16gbsim-ncpu18-cap13g-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 59.381116758629034,
      "decode": 16.725064012361415,
      "ttft": null,
      "latency": 97.25,
      "wall": 194.5,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 194.5,
        "median_decode_tps": 16.725,
        "median_prefill_tps": 59.381,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 33.45012802472283,
            "min": 16.652336845268014,
            "median": 16.725064012361415,
            "mean": 16.725064012361415,
            "p95": 16.79779117945482,
            "max": 16.79779117945482
          },
          "prefill_tps": {
            "n": 2,
            "total": 118.76223351725807,
            "min": 51.848596912156815,
            "median": 59.381116758629034,
            "mean": 59.381116758629034,
            "p95": 66.91363660510126,
            "max": 66.91363660510126
          },
          "elapsed_s": {
            "n": 2,
            "total": 194.5,
            "min": 95,
            "median": 97.25,
            "mean": 97.25,
            "p95": 99.5,
            "max": 99.5
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2136,
            "min": 927,
            "median": 1068,
            "mean": 1068,
            "p95": 1209,
            "max": 1209
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 95,
          "median_decode_tps": 16.652,
          "median_prefill_tps": 51.849,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 16.652336845268014,
              "min": 16.652336845268014,
              "median": 16.652336845268014,
              "mean": 16.652336845268014,
              "p95": 16.652336845268014,
              "max": 16.652336845268014
            },
            "prefill_tps": {
              "n": 1,
              "total": 51.848596912156815,
              "min": 51.848596912156815,
              "median": 51.848596912156815,
              "mean": 51.848596912156815,
              "p95": 51.848596912156815,
              "max": 51.848596912156815
            },
            "elapsed_s": {
              "n": 1,
              "total": 95,
              "min": 95,
              "median": 95,
              "mean": 95,
              "p95": 95,
              "max": 95
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 927,
              "min": 927,
              "median": 927,
              "mean": 927,
              "p95": 927,
              "max": 927
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 99.5,
          "median_decode_tps": 16.798,
          "median_prefill_tps": 66.914,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 16.79779117945482,
              "min": 16.79779117945482,
              "median": 16.79779117945482,
              "mean": 16.79779117945482,
              "p95": 16.79779117945482,
              "max": 16.79779117945482
            },
            "prefill_tps": {
              "n": 1,
              "total": 66.91363660510126,
              "min": 66.91363660510126,
              "median": 66.91363660510126,
              "mean": 66.91363660510126,
              "p95": 66.91363660510126,
              "max": 66.91363660510126
            },
            "elapsed_s": {
              "n": 1,
              "total": 99.5,
              "min": 99.5,
              "median": 99.5,
              "mean": 99.5,
              "p95": 99.5,
              "max": 99.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1209,
              "min": 1209,
              "median": 1209,
              "mean": 1209,
              "p95": 1209,
              "max": 1209
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-17",
      "run": "knobs/b70-moe-16gbsim-ncpu18-cap13g-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 62.22658482082399,
      "decode": 15.959576799361887,
      "ttft": null,
      "latency": 97.2,
      "wall": 194.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 194.4,
        "median_decode_tps": 15.96,
        "median_prefill_tps": 62.227,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 31.919153598723774,
            "min": 15.042876797131106,
            "median": 15.959576799361887,
            "mean": 15.959576799361887,
            "p95": 16.876276801592667,
            "max": 16.876276801592667
          },
          "prefill_tps": {
            "n": 2,
            "total": 124.45316964164797,
            "min": 59.16730029817999,
            "median": 62.22658482082399,
            "mean": 62.22658482082399,
            "p95": 65.285869343468,
            "max": 65.285869343468
          },
          "elapsed_s": {
            "n": 2,
            "total": 194.4,
            "min": 95.7,
            "median": 97.2,
            "mean": 97.2,
            "p95": 98.7,
            "max": 98.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2105,
            "min": 917,
            "median": 1052.5,
            "mean": 1052.5,
            "p95": 1188,
            "max": 1188
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 95.7,
          "median_decode_tps": 15.043,
          "median_prefill_tps": 59.167,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.042876797131106,
              "min": 15.042876797131106,
              "median": 15.042876797131106,
              "mean": 15.042876797131106,
              "p95": 15.042876797131106,
              "max": 15.042876797131106
            },
            "prefill_tps": {
              "n": 1,
              "total": 59.16730029817999,
              "min": 59.16730029817999,
              "median": 59.16730029817999,
              "mean": 59.16730029817999,
              "p95": 59.16730029817999,
              "max": 59.16730029817999
            },
            "elapsed_s": {
              "n": 1,
              "total": 95.7,
              "min": 95.7,
              "median": 95.7,
              "mean": 95.7,
              "p95": 95.7,
              "max": 95.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 917,
              "min": 917,
              "median": 917,
              "mean": 917,
              "p95": 917,
              "max": 917
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 98.7,
          "median_decode_tps": 16.876,
          "median_prefill_tps": 65.286,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 16.876276801592667,
              "min": 16.876276801592667,
              "median": 16.876276801592667,
              "mean": 16.876276801592667,
              "p95": 16.876276801592667,
              "max": 16.876276801592667
            },
            "prefill_tps": {
              "n": 1,
              "total": 65.285869343468,
              "min": 65.285869343468,
              "median": 65.285869343468,
              "mean": 65.285869343468,
              "p95": 65.285869343468,
              "max": 65.285869343468
            },
            "elapsed_s": {
              "n": 1,
              "total": 98.7,
              "min": 98.7,
              "median": 98.7,
              "mean": 98.7,
              "p95": 98.7,
              "max": 98.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1188,
              "min": 1188,
              "median": 1188,
              "mean": 1188,
              "p95": 1188,
              "max": 1188
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-18",
      "run": "knobs/b70-moe-cap30g-ncpu18-all24-p1.jsonl",
      "model": "Flash-Next MoE Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "SYCL b10964, 30 expert layers on the card, experts memory-mapped from NVMe (`--load-mode mmap --no-host --no-repack`), 32k, 1 slot, 30 GiB container cap, request `repeat_penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 92.969,
      "structure": 23,
      "strict": null,
      "prefill": 105.47638190555719,
      "decode": 20.361416398509093,
      "ttft": null,
      "latency": 81.9,
      "wall": 2100.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": 32768,
      "slots": 1,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 119,
        "checks_possible": 128,
        "check_percent": 92.969,
        "wall_s": 2100.8,
        "median_decode_tps": 20.361,
        "median_prefill_tps": 105.476,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 487.4997993954328,
            "min": 19.804485334198237,
            "median": 20.361416398509093,
            "mean": 20.312491641476367,
            "p95": 20.6855758587917,
            "max": 20.800276101421208
          },
          "prefill_tps": {
            "n": 24,
            "total": 2529.2790013637077,
            "min": 83.12207618097034,
            "median": 105.47638190555719,
            "mean": 105.38662505682116,
            "p95": 111.32519803040036,
            "max": 111.7920967868069
          },
          "elapsed_s": {
            "n": 24,
            "total": 2100.8,
            "min": 47.2,
            "median": 81.9,
            "mean": 87.53333333333333,
            "p95": 131.6,
            "max": 146.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 30680,
            "min": 723,
            "median": 1201,
            "mean": 1278.3333333333333,
            "p95": 1952,
            "max": 2108
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 63,
          "checks_possible": 65,
          "check_percent": 96.923,
          "wall_s": 946.5,
          "median_decode_tps": 20.381,
          "median_prefill_tps": 104.194,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 243.78714801058663,
              "min": 19.804485334198237,
              "median": 20.380822047893112,
              "mean": 20.315595667548887,
              "p95": 20.638317431728904,
              "max": 20.638317431728904
            },
            "prefill_tps": {
              "n": 12,
              "total": 1229.2631886733272,
              "min": 83.12207618097034,
              "median": 104.19371335850326,
              "mean": 102.4385990561106,
              "p95": 106.4226446382998,
              "max": 106.4226446382998
            },
            "elapsed_s": {
              "n": 12,
              "total": 946.5,
              "min": 47.2,
              "median": 66.69999999999999,
              "mean": 78.875,
              "p95": 146.3,
              "max": 146.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13941,
              "min": 723,
              "median": 997,
              "mean": 1161.75,
              "p95": 2108,
              "max": 2108
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 56,
          "checks_possible": 63,
          "check_percent": 88.889,
          "wall_s": 1154.3,
          "median_decode_tps": 20.317,
          "median_prefill_tps": 107.723,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 243.71265138484617,
              "min": 19.913864116916326,
              "median": 20.317361714907094,
              "mean": 20.309387615403846,
              "p95": 20.800276101421208,
              "max": 20.800276101421208
            },
            "prefill_tps": {
              "n": 12,
              "total": 1300.0158126903807,
              "min": 104.47701601747613,
              "median": 107.72316976143999,
              "mean": 108.33465105753172,
              "p95": 111.7920967868069,
              "max": 111.7920967868069
            },
            "elapsed_s": {
              "n": 12,
              "total": 1154.3,
              "min": 76.3,
              "median": 88.6,
              "mean": 96.19166666666666,
              "p95": 131.6,
              "max": 131.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 16739,
              "min": 1045,
              "median": 1321.5,
              "mean": 1394.9166666666667,
              "p95": 1952,
              "max": 1952
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-19",
      "run": "knobs/b70-moe-hdd-auto-ncpu40-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 13.939861861031224,
      "decode": 11.090698331887065,
      "ttft": null,
      "latency": 235.3,
      "wall": 470.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 470.6,
        "median_decode_tps": 11.091,
        "median_prefill_tps": 13.94,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 22.18139666377413,
            "min": 10.562071937170291,
            "median": 11.090698331887065,
            "mean": 11.090698331887065,
            "p95": 11.619324726603837,
            "max": 11.619324726603837
          },
          "prefill_tps": {
            "n": 2,
            "total": 27.879723722062447,
            "min": 12.63041410465477,
            "median": 13.939861861031224,
            "mean": 13.939861861031224,
            "p95": 15.249309617407677,
            "max": 15.249309617407677
          },
          "elapsed_s": {
            "n": 2,
            "total": 470.6,
            "min": 210.9,
            "median": 235.3,
            "mean": 235.3,
            "p95": 259.7,
            "max": 259.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2106,
            "min": 897,
            "median": 1053,
            "mean": 1053,
            "p95": 1209,
            "max": 1209
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 210.9,
          "median_decode_tps": 11.619,
          "median_prefill_tps": 15.249,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.619324726603837,
              "min": 11.619324726603837,
              "median": 11.619324726603837,
              "mean": 11.619324726603837,
              "p95": 11.619324726603837,
              "max": 11.619324726603837
            },
            "prefill_tps": {
              "n": 1,
              "total": 15.249309617407677,
              "min": 15.249309617407677,
              "median": 15.249309617407677,
              "mean": 15.249309617407677,
              "p95": 15.249309617407677,
              "max": 15.249309617407677
            },
            "elapsed_s": {
              "n": 1,
              "total": 210.9,
              "min": 210.9,
              "median": 210.9,
              "mean": 210.9,
              "p95": 210.9,
              "max": 210.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 897,
              "min": 897,
              "median": 897,
              "mean": 897,
              "p95": 897,
              "max": 897
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 259.7,
          "median_decode_tps": 10.562,
          "median_prefill_tps": 12.63,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 10.562071937170291,
              "min": 10.562071937170291,
              "median": 10.562071937170291,
              "mean": 10.562071937170291,
              "p95": 10.562071937170291,
              "max": 10.562071937170291
            },
            "prefill_tps": {
              "n": 1,
              "total": 12.63041410465477,
              "min": 12.63041410465477,
              "median": 12.63041410465477,
              "mean": 12.63041410465477,
              "p95": 12.63041410465477,
              "max": 12.63041410465477
            },
            "elapsed_s": {
              "n": 1,
              "total": 259.7,
              "min": 259.7,
              "median": 259.7,
              "mean": 259.7,
              "p95": 259.7,
              "max": 259.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1209,
              "min": 1209,
              "median": 1209,
              "mean": 1209,
              "p95": 1209,
              "max": 1209
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-20",
      "run": "knobs/b70-moe-hdd-auto-ncpu40-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 63.684811679552084,
      "decode": 13.899715019140551,
      "ttft": null,
      "latency": 102.65,
      "wall": 205.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 205.3,
        "median_decode_tps": 13.9,
        "median_prefill_tps": 63.685,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 27.799430038281102,
            "min": 13.471687076755444,
            "median": 13.899715019140551,
            "mean": 13.899715019140551,
            "p95": 14.32774296152566,
            "max": 14.32774296152566
          },
          "prefill_tps": {
            "n": 2,
            "total": 127.36962335910417,
            "min": 63.11300048609385,
            "median": 63.684811679552084,
            "mean": 63.684811679552084,
            "p95": 64.25662287301031,
            "max": 64.25662287301031
          },
          "elapsed_s": {
            "n": 2,
            "total": 205.3,
            "min": 100.6,
            "median": 102.65,
            "mean": 102.65,
            "p95": 104.7,
            "max": 104.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2000,
            "min": 976,
            "median": 1000,
            "mean": 1000,
            "p95": 1024,
            "max": 1024
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 100.6,
          "median_decode_tps": 14.328,
          "median_prefill_tps": 63.113,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.32774296152566,
              "min": 14.32774296152566,
              "median": 14.32774296152566,
              "mean": 14.32774296152566,
              "p95": 14.32774296152566,
              "max": 14.32774296152566
            },
            "prefill_tps": {
              "n": 1,
              "total": 63.11300048609385,
              "min": 63.11300048609385,
              "median": 63.11300048609385,
              "mean": 63.11300048609385,
              "p95": 63.11300048609385,
              "max": 63.11300048609385
            },
            "elapsed_s": {
              "n": 1,
              "total": 100.6,
              "min": 100.6,
              "median": 100.6,
              "mean": 100.6,
              "p95": 100.6,
              "max": 100.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 976,
              "min": 976,
              "median": 976,
              "mean": 976,
              "p95": 976,
              "max": 976
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 104.7,
          "median_decode_tps": 13.472,
          "median_prefill_tps": 64.257,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 13.471687076755444,
              "min": 13.471687076755444,
              "median": 13.471687076755444,
              "mean": 13.471687076755444,
              "p95": 13.471687076755444,
              "max": 13.471687076755444
            },
            "prefill_tps": {
              "n": 1,
              "total": 64.25662287301031,
              "min": 64.25662287301031,
              "median": 64.25662287301031,
              "mean": 64.25662287301031,
              "p95": 64.25662287301031,
              "max": 64.25662287301031
            },
            "elapsed_s": {
              "n": 1,
              "total": 104.7,
              "min": 104.7,
              "median": 104.7,
              "mean": 104.7,
              "p95": 104.7,
              "max": 104.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1024,
              "min": 1024,
              "median": 1024,
              "mean": 1024,
              "p95": 1024,
              "max": 1024
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-21",
      "run": "knobs/b70-moe-mmap-nohost-norepack-ncpu20-cap10g-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 45.031941892510396,
      "decode": 11.633103577087397,
      "ttft": null,
      "latency": 138.05,
      "wall": 276.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 276.1,
        "median_decode_tps": 11.633,
        "median_prefill_tps": 45.032,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 23.266207154174793,
            "min": 11.614742136190104,
            "median": 11.633103577087397,
            "mean": 11.633103577087397,
            "p95": 11.65146501798469,
            "max": 11.65146501798469
          },
          "prefill_tps": {
            "n": 2,
            "total": 90.06388378502079,
            "min": 43.97056058886754,
            "median": 45.031941892510396,
            "mean": 45.031941892510396,
            "p95": 46.09332319615325,
            "max": 46.09332319615325
          },
          "elapsed_s": {
            "n": 2,
            "total": 276.1,
            "min": 132.6,
            "median": 138.05,
            "mean": 138.05,
            "p95": 143.5,
            "max": 143.5
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2208,
            "min": 1002,
            "median": 1104,
            "mean": 1104,
            "p95": 1206,
            "max": 1206
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 132.6,
          "median_decode_tps": 11.615,
          "median_prefill_tps": 43.971,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.614742136190104,
              "min": 11.614742136190104,
              "median": 11.614742136190104,
              "mean": 11.614742136190104,
              "p95": 11.614742136190104,
              "max": 11.614742136190104
            },
            "prefill_tps": {
              "n": 1,
              "total": 43.97056058886754,
              "min": 43.97056058886754,
              "median": 43.97056058886754,
              "mean": 43.97056058886754,
              "p95": 43.97056058886754,
              "max": 43.97056058886754
            },
            "elapsed_s": {
              "n": 1,
              "total": 132.6,
              "min": 132.6,
              "median": 132.6,
              "mean": 132.6,
              "p95": 132.6,
              "max": 132.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 1002,
              "min": 1002,
              "median": 1002,
              "mean": 1002,
              "p95": 1002,
              "max": 1002
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 143.5,
          "median_decode_tps": 11.651,
          "median_prefill_tps": 46.093,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.65146501798469,
              "min": 11.65146501798469,
              "median": 11.65146501798469,
              "mean": 11.65146501798469,
              "p95": 11.65146501798469,
              "max": 11.65146501798469
            },
            "prefill_tps": {
              "n": 1,
              "total": 46.09332319615325,
              "min": 46.09332319615325,
              "median": 46.09332319615325,
              "mean": 46.09332319615325,
              "p95": 46.09332319615325,
              "max": 46.09332319615325
            },
            "elapsed_s": {
              "n": 1,
              "total": 143.5,
              "min": 143.5,
              "median": 143.5,
              "mean": 143.5,
              "p95": 143.5,
              "max": 143.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1206,
              "min": 1206,
              "median": 1206,
              "mean": 1206,
              "p95": 1206,
              "max": 1206
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-22",
      "run": "knobs/b70-moe-mmap-nohost-norepack-ncpu20-cap10g-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 45.564191204608925,
      "decode": 11.598486571442898,
      "ttft": null,
      "latency": 130.45,
      "wall": 260.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 260.9,
        "median_decode_tps": 11.598,
        "median_prefill_tps": 45.564,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 23.196973142885795,
            "min": 11.489387163549338,
            "median": 11.598486571442898,
            "mean": 11.598486571442898,
            "p95": 11.707585979336457,
            "max": 11.707585979336457
          },
          "prefill_tps": {
            "n": 2,
            "total": 91.12838240921785,
            "min": 45.53745853300988,
            "median": 45.564191204608925,
            "mean": 45.564191204608925,
            "p95": 45.59092387620797,
            "max": 45.59092387620797
          },
          "elapsed_s": {
            "n": 2,
            "total": 260.9,
            "min": 125.5,
            "median": 130.45,
            "mean": 130.45,
            "p95": 135.4,
            "max": 135.4
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2037,
            "min": 925,
            "median": 1018.5,
            "mean": 1018.5,
            "p95": 1112,
            "max": 1112
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 125.5,
          "median_decode_tps": 11.489,
          "median_prefill_tps": 45.537,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.489387163549338,
              "min": 11.489387163549338,
              "median": 11.489387163549338,
              "mean": 11.489387163549338,
              "p95": 11.489387163549338,
              "max": 11.489387163549338
            },
            "prefill_tps": {
              "n": 1,
              "total": 45.53745853300988,
              "min": 45.53745853300988,
              "median": 45.53745853300988,
              "mean": 45.53745853300988,
              "p95": 45.53745853300988,
              "max": 45.53745853300988
            },
            "elapsed_s": {
              "n": 1,
              "total": 125.5,
              "min": 125.5,
              "median": 125.5,
              "mean": 125.5,
              "p95": 125.5,
              "max": 125.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 925,
              "min": 925,
              "median": 925,
              "mean": 925,
              "p95": 925,
              "max": 925
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 135.4,
          "median_decode_tps": 11.708,
          "median_prefill_tps": 45.591,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.707585979336457,
              "min": 11.707585979336457,
              "median": 11.707585979336457,
              "mean": 11.707585979336457,
              "p95": 11.707585979336457,
              "max": 11.707585979336457
            },
            "prefill_tps": {
              "n": 1,
              "total": 45.59092387620797,
              "min": 45.59092387620797,
              "median": 45.59092387620797,
              "mean": 45.59092387620797,
              "p95": 45.59092387620797,
              "max": 45.59092387620797
            },
            "elapsed_s": {
              "n": 1,
              "total": 135.4,
              "min": 135.4,
              "median": 135.4,
              "mean": 135.4,
              "p95": 135.4,
              "max": 135.4
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1112,
              "min": 1112,
              "median": 1112,
              "mean": 1112,
              "p95": 1112,
              "max": 1112
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-23",
      "run": "knobs/b70-moe-mmap-nohost-norepack-ncpu20-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 83.62357705304382,
      "decode": 19.486259956503183,
      "ttft": null,
      "latency": 80.55,
      "wall": 161.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 161.1,
        "median_decode_tps": 19.486,
        "median_prefill_tps": 83.624,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 38.972519913006366,
            "min": 19.40239836926722,
            "median": 19.486259956503183,
            "mean": 19.486259956503183,
            "p95": 19.57012154373914,
            "max": 19.57012154373914
          },
          "prefill_tps": {
            "n": 2,
            "total": 167.24715410608763,
            "min": 71.7937744943466,
            "median": 83.62357705304382,
            "mean": 83.62357705304382,
            "p95": 95.45337961174104,
            "max": 95.45337961174104
          },
          "elapsed_s": {
            "n": 2,
            "total": 161.1,
            "min": 80,
            "median": 80.55,
            "mean": 80.55,
            "p95": 81.1,
            "max": 81.1
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2210,
            "min": 1001,
            "median": 1105,
            "mean": 1105,
            "p95": 1209,
            "max": 1209
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 80,
          "median_decode_tps": 19.402,
          "median_prefill_tps": 71.794,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.40239836926722,
              "min": 19.40239836926722,
              "median": 19.40239836926722,
              "mean": 19.40239836926722,
              "p95": 19.40239836926722,
              "max": 19.40239836926722
            },
            "prefill_tps": {
              "n": 1,
              "total": 71.7937744943466,
              "min": 71.7937744943466,
              "median": 71.7937744943466,
              "mean": 71.7937744943466,
              "p95": 71.7937744943466,
              "max": 71.7937744943466
            },
            "elapsed_s": {
              "n": 1,
              "total": 80,
              "min": 80,
              "median": 80,
              "mean": 80,
              "p95": 80,
              "max": 80
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 1001,
              "min": 1001,
              "median": 1001,
              "mean": 1001,
              "p95": 1001,
              "max": 1001
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 81.1,
          "median_decode_tps": 19.57,
          "median_prefill_tps": 95.453,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.57012154373914,
              "min": 19.57012154373914,
              "median": 19.57012154373914,
              "mean": 19.57012154373914,
              "p95": 19.57012154373914,
              "max": 19.57012154373914
            },
            "prefill_tps": {
              "n": 1,
              "total": 95.45337961174104,
              "min": 95.45337961174104,
              "median": 95.45337961174104,
              "mean": 95.45337961174104,
              "p95": 95.45337961174104,
              "max": 95.45337961174104
            },
            "elapsed_s": {
              "n": 1,
              "total": 81.1,
              "min": 81.1,
              "median": 81.1,
              "mean": 81.1,
              "p95": 81.1,
              "max": 81.1
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1209,
              "min": 1209,
              "median": 1209,
              "mean": 1209,
              "p95": 1209,
              "max": 1209
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-24",
      "run": "knobs/b70-moe-mmap-nohost-norepack-ncpu20-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 102.90802564726802,
      "decode": 19.788433183345205,
      "ttft": null,
      "latency": 70.65,
      "wall": 141.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 141.3,
        "median_decode_tps": 19.788,
        "median_prefill_tps": 102.908,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 39.57686636669041,
            "min": 19.737997140470263,
            "median": 19.788433183345205,
            "mean": 19.788433183345205,
            "p95": 19.838869226220147,
            "max": 19.838869226220147
          },
          "prefill_tps": {
            "n": 2,
            "total": 205.81605129453604,
            "min": 101.75472024684498,
            "median": 102.90802564726802,
            "mean": 102.90802564726802,
            "p95": 104.06133104769106,
            "max": 104.06133104769106
          },
          "elapsed_s": {
            "n": 2,
            "total": 141.3,
            "min": 69.8,
            "median": 70.65,
            "mean": 70.65,
            "p95": 71.5,
            "max": 71.5
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2045,
            "min": 980,
            "median": 1022.5,
            "mean": 1022.5,
            "p95": 1065,
            "max": 1065
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 69.8,
          "median_decode_tps": 19.738,
          "median_prefill_tps": 101.755,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.737997140470263,
              "min": 19.737997140470263,
              "median": 19.737997140470263,
              "mean": 19.737997140470263,
              "p95": 19.737997140470263,
              "max": 19.737997140470263
            },
            "prefill_tps": {
              "n": 1,
              "total": 101.75472024684498,
              "min": 101.75472024684498,
              "median": 101.75472024684498,
              "mean": 101.75472024684498,
              "p95": 101.75472024684498,
              "max": 101.75472024684498
            },
            "elapsed_s": {
              "n": 1,
              "total": 69.8,
              "min": 69.8,
              "median": 69.8,
              "mean": 69.8,
              "p95": 69.8,
              "max": 69.8
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 980,
              "min": 980,
              "median": 980,
              "mean": 980,
              "p95": 980,
              "max": 980
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 71.5,
          "median_decode_tps": 19.839,
          "median_prefill_tps": 104.061,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.838869226220147,
              "min": 19.838869226220147,
              "median": 19.838869226220147,
              "mean": 19.838869226220147,
              "p95": 19.838869226220147,
              "max": 19.838869226220147
            },
            "prefill_tps": {
              "n": 1,
              "total": 104.06133104769106,
              "min": 104.06133104769106,
              "median": 104.06133104769106,
              "mean": 104.06133104769106,
              "p95": 104.06133104769106,
              "max": 104.06133104769106
            },
            "elapsed_s": {
              "n": 1,
              "total": 71.5,
              "min": 71.5,
              "median": 71.5,
              "mean": 71.5,
              "p95": 71.5,
              "max": 71.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1065,
              "min": 1065,
              "median": 1065,
              "mean": 1065,
              "p95": 1065,
              "max": 1065
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-25",
      "run": "knobs/b70-moe-nvme-auto-ncpu20-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 96.61072162282687,
      "decode": 19.901418385345178,
      "ttft": null,
      "latency": 72.2,
      "wall": 144.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 144.4,
        "median_decode_tps": 19.901,
        "median_prefill_tps": 96.611,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 39.802836770690355,
            "min": 19.883546925138912,
            "median": 19.901418385345178,
            "mean": 19.901418385345178,
            "p95": 19.919289845551443,
            "max": 19.919289845551443
          },
          "prefill_tps": {
            "n": 2,
            "total": 193.22144324565375,
            "min": 94.64301512975302,
            "median": 96.61072162282687,
            "mean": 96.61072162282687,
            "p95": 98.57842811590072,
            "max": 98.57842811590072
          },
          "elapsed_s": {
            "n": 2,
            "total": 144.4,
            "min": 68.7,
            "median": 72.2,
            "mean": 72.2,
            "p95": 75.7,
            "max": 75.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2074,
            "min": 938,
            "median": 1037,
            "mean": 1037,
            "p95": 1136,
            "max": 1136
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 68.7,
          "median_decode_tps": 19.884,
          "median_prefill_tps": 94.643,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.883546925138912,
              "min": 19.883546925138912,
              "median": 19.883546925138912,
              "mean": 19.883546925138912,
              "p95": 19.883546925138912,
              "max": 19.883546925138912
            },
            "prefill_tps": {
              "n": 1,
              "total": 94.64301512975302,
              "min": 94.64301512975302,
              "median": 94.64301512975302,
              "mean": 94.64301512975302,
              "p95": 94.64301512975302,
              "max": 94.64301512975302
            },
            "elapsed_s": {
              "n": 1,
              "total": 68.7,
              "min": 68.7,
              "median": 68.7,
              "mean": 68.7,
              "p95": 68.7,
              "max": 68.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 938,
              "min": 938,
              "median": 938,
              "mean": 938,
              "p95": 938,
              "max": 938
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 75.7,
          "median_decode_tps": 19.919,
          "median_prefill_tps": 98.578,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 19.919289845551443,
              "min": 19.919289845551443,
              "median": 19.919289845551443,
              "mean": 19.919289845551443,
              "p95": 19.919289845551443,
              "max": 19.919289845551443
            },
            "prefill_tps": {
              "n": 1,
              "total": 98.57842811590072,
              "min": 98.57842811590072,
              "median": 98.57842811590072,
              "mean": 98.57842811590072,
              "p95": 98.57842811590072,
              "max": 98.57842811590072
            },
            "elapsed_s": {
              "n": 1,
              "total": 75.7,
              "min": 75.7,
              "median": 75.7,
              "mean": 75.7,
              "p95": 75.7,
              "max": 75.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1136,
              "min": 1136,
              "median": 1136,
              "mean": 1136,
              "p95": 1136,
              "max": 1136
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-26",
      "run": "knobs/b70-moe-nvme-auto-ncpu20-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 104.02136084899752,
      "decode": 20.20014872747614,
      "ttft": null,
      "latency": 71.35,
      "wall": 142.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 142.7,
        "median_decode_tps": 20.2,
        "median_prefill_tps": 104.021,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 40.40029745495228,
            "min": 20.181366603896645,
            "median": 20.20014872747614,
            "mean": 20.20014872747614,
            "p95": 20.21893085105564,
            "max": 20.21893085105564
          },
          "prefill_tps": {
            "n": 2,
            "total": 208.04272169799503,
            "min": 103.46838713630291,
            "median": 104.02136084899752,
            "mean": 104.02136084899752,
            "p95": 104.57433456169214,
            "max": 104.57433456169214
          },
          "elapsed_s": {
            "n": 2,
            "total": 142.7,
            "min": 65.2,
            "median": 71.35,
            "mean": 71.35,
            "p95": 77.5,
            "max": 77.5
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2123,
            "min": 914,
            "median": 1061.5,
            "mean": 1061.5,
            "p95": 1209,
            "max": 1209
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 65.2,
          "median_decode_tps": 20.181,
          "median_prefill_tps": 103.468,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 20.181366603896645,
              "min": 20.181366603896645,
              "median": 20.181366603896645,
              "mean": 20.181366603896645,
              "p95": 20.181366603896645,
              "max": 20.181366603896645
            },
            "prefill_tps": {
              "n": 1,
              "total": 103.46838713630291,
              "min": 103.46838713630291,
              "median": 103.46838713630291,
              "mean": 103.46838713630291,
              "p95": 103.46838713630291,
              "max": 103.46838713630291
            },
            "elapsed_s": {
              "n": 1,
              "total": 65.2,
              "min": 65.2,
              "median": 65.2,
              "mean": 65.2,
              "p95": 65.2,
              "max": 65.2
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 914,
              "min": 914,
              "median": 914,
              "mean": 914,
              "p95": 914,
              "max": 914
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 77.5,
          "median_decode_tps": 20.219,
          "median_prefill_tps": 104.574,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 20.21893085105564,
              "min": 20.21893085105564,
              "median": 20.21893085105564,
              "mean": 20.21893085105564,
              "p95": 20.21893085105564,
              "max": 20.21893085105564
            },
            "prefill_tps": {
              "n": 1,
              "total": 104.57433456169214,
              "min": 104.57433456169214,
              "median": 104.57433456169214,
              "mean": 104.57433456169214,
              "p95": 104.57433456169214,
              "max": 104.57433456169214
            },
            "elapsed_s": {
              "n": 1,
              "total": 77.5,
              "min": 77.5,
              "median": 77.5,
              "mean": 77.5,
              "p95": 77.5,
              "max": 77.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1209,
              "min": 1209,
              "median": 1209,
              "mean": 1209,
              "p95": 1209,
              "max": 1209
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-27",
      "run": "knobs/b70-moe-nvme-auto-ncpu40-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 65.12854883569767,
      "decode": 15.673086747465131,
      "ttft": null,
      "latency": 98.05,
      "wall": 196.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 196.1,
        "median_decode_tps": 15.673,
        "median_prefill_tps": 65.129,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 31.346173494930262,
            "min": 15.656369539892761,
            "median": 15.673086747465131,
            "mean": 15.673086747465131,
            "p95": 15.689803955037503,
            "max": 15.689803955037503
          },
          "prefill_tps": {
            "n": 2,
            "total": 130.25709767139534,
            "min": 64.45470748747587,
            "median": 65.12854883569767,
            "mean": 65.12854883569767,
            "p95": 65.80239018391947,
            "max": 65.80239018391947
          },
          "elapsed_s": {
            "n": 2,
            "total": 196.1,
            "min": 95.5,
            "median": 98.05,
            "mean": 98.05,
            "p95": 100.6,
            "max": 100.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2140,
            "min": 1002,
            "median": 1070,
            "mean": 1070,
            "p95": 1138,
            "max": 1138
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 95.5,
          "median_decode_tps": 15.69,
          "median_prefill_tps": 64.455,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.689803955037503,
              "min": 15.689803955037503,
              "median": 15.689803955037503,
              "mean": 15.689803955037503,
              "p95": 15.689803955037503,
              "max": 15.689803955037503
            },
            "prefill_tps": {
              "n": 1,
              "total": 64.45470748747587,
              "min": 64.45470748747587,
              "median": 64.45470748747587,
              "mean": 64.45470748747587,
              "p95": 64.45470748747587,
              "max": 64.45470748747587
            },
            "elapsed_s": {
              "n": 1,
              "total": 95.5,
              "min": 95.5,
              "median": 95.5,
              "mean": 95.5,
              "p95": 95.5,
              "max": 95.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 1002,
              "min": 1002,
              "median": 1002,
              "mean": 1002,
              "p95": 1002,
              "max": 1002
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 100.6,
          "median_decode_tps": 15.656,
          "median_prefill_tps": 65.802,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.656369539892761,
              "min": 15.656369539892761,
              "median": 15.656369539892761,
              "mean": 15.656369539892761,
              "p95": 15.656369539892761,
              "max": 15.656369539892761
            },
            "prefill_tps": {
              "n": 1,
              "total": 65.80239018391947,
              "min": 65.80239018391947,
              "median": 65.80239018391947,
              "mean": 65.80239018391947,
              "p95": 65.80239018391947,
              "max": 65.80239018391947
            },
            "elapsed_s": {
              "n": 1,
              "total": 100.6,
              "min": 100.6,
              "median": 100.6,
              "mean": 100.6,
              "p95": 100.6,
              "max": 100.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1138,
              "min": 1138,
              "median": 1138,
              "mean": 1138,
              "p95": 1138,
              "max": 1138
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-28",
      "run": "knobs/b70-moe-nvme-auto-ncpu40-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 68.06699756467583,
      "decode": 15.44240007782042,
      "ttft": null,
      "latency": 98.3,
      "wall": 196.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 196.6,
        "median_decode_tps": 15.442,
        "median_prefill_tps": 68.067,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 30.88480015564084,
            "min": 15.32994634095128,
            "median": 15.44240007782042,
            "mean": 15.44240007782042,
            "p95": 15.554853814689558,
            "max": 15.554853814689558
          },
          "prefill_tps": {
            "n": 2,
            "total": 136.13399512935166,
            "min": 67.8226031990725,
            "median": 68.06699756467583,
            "mean": 68.06699756467583,
            "p95": 68.31139193027917,
            "max": 68.31139193027917
          },
          "elapsed_s": {
            "n": 2,
            "total": 196.6,
            "min": 94,
            "median": 98.3,
            "mean": 98.3,
            "p95": 102.6,
            "max": 102.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2154,
            "min": 978,
            "median": 1077,
            "mean": 1077,
            "p95": 1176,
            "max": 1176
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 94,
          "median_decode_tps": 15.33,
          "median_prefill_tps": 67.823,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.32994634095128,
              "min": 15.32994634095128,
              "median": 15.32994634095128,
              "mean": 15.32994634095128,
              "p95": 15.32994634095128,
              "max": 15.32994634095128
            },
            "prefill_tps": {
              "n": 1,
              "total": 67.8226031990725,
              "min": 67.8226031990725,
              "median": 67.8226031990725,
              "mean": 67.8226031990725,
              "p95": 67.8226031990725,
              "max": 67.8226031990725
            },
            "elapsed_s": {
              "n": 1,
              "total": 94,
              "min": 94,
              "median": 94,
              "mean": 94,
              "p95": 94,
              "max": 94
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 978,
              "min": 978,
              "median": 978,
              "mean": 978,
              "p95": 978,
              "max": 978
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 102.6,
          "median_decode_tps": 15.555,
          "median_prefill_tps": 68.311,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.554853814689558,
              "min": 15.554853814689558,
              "median": 15.554853814689558,
              "mean": 15.554853814689558,
              "p95": 15.554853814689558,
              "max": 15.554853814689558
            },
            "prefill_tps": {
              "n": 1,
              "total": 68.31139193027917,
              "min": 68.31139193027917,
              "median": 68.31139193027917,
              "mean": 68.31139193027917,
              "p95": 68.31139193027917,
              "max": 68.31139193027917
            },
            "elapsed_s": {
              "n": 1,
              "total": 102.6,
              "min": 102.6,
              "median": 102.6,
              "mean": 102.6,
              "p95": 102.6,
              "max": 102.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1176,
              "min": 1176,
              "median": 1176,
              "mean": 1176,
              "p95": 1176,
              "max": 1176
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-29",
      "run": "knobs/b70-moe-nvme-auto-ncpu48-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 56.61186817127191,
      "decode": 14.078387139294936,
      "ttft": null,
      "latency": 108.35,
      "wall": 216.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 216.7,
        "median_decode_tps": 14.078,
        "median_prefill_tps": 56.612,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 28.15677427858987,
            "min": 14.027160590663692,
            "median": 14.078387139294936,
            "mean": 14.078387139294936,
            "p95": 14.12961368792618,
            "max": 14.12961368792618
          },
          "prefill_tps": {
            "n": 2,
            "total": 113.22373634254382,
            "min": 55.31609051252581,
            "median": 56.61186817127191,
            "mean": 56.61186817127191,
            "p95": 57.90764583001801,
            "max": 57.90764583001801
          },
          "elapsed_s": {
            "n": 2,
            "total": 216.7,
            "min": 108,
            "median": 108.35,
            "mean": 108.35,
            "p95": 108.7,
            "max": 108.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2085,
            "min": 1006,
            "median": 1042.5,
            "mean": 1042.5,
            "p95": 1079,
            "max": 1079
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 108,
          "median_decode_tps": 14.13,
          "median_prefill_tps": 55.316,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.12961368792618,
              "min": 14.12961368792618,
              "median": 14.12961368792618,
              "mean": 14.12961368792618,
              "p95": 14.12961368792618,
              "max": 14.12961368792618
            },
            "prefill_tps": {
              "n": 1,
              "total": 55.31609051252581,
              "min": 55.31609051252581,
              "median": 55.31609051252581,
              "mean": 55.31609051252581,
              "p95": 55.31609051252581,
              "max": 55.31609051252581
            },
            "elapsed_s": {
              "n": 1,
              "total": 108,
              "min": 108,
              "median": 108,
              "mean": 108,
              "p95": 108,
              "max": 108
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 1006,
              "min": 1006,
              "median": 1006,
              "mean": 1006,
              "p95": 1006,
              "max": 1006
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 108.7,
          "median_decode_tps": 14.027,
          "median_prefill_tps": 57.908,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.027160590663692,
              "min": 14.027160590663692,
              "median": 14.027160590663692,
              "mean": 14.027160590663692,
              "p95": 14.027160590663692,
              "max": 14.027160590663692
            },
            "prefill_tps": {
              "n": 1,
              "total": 57.90764583001801,
              "min": 57.90764583001801,
              "median": 57.90764583001801,
              "mean": 57.90764583001801,
              "p95": 57.90764583001801,
              "max": 57.90764583001801
            },
            "elapsed_s": {
              "n": 1,
              "total": 108.7,
              "min": 108.7,
              "median": 108.7,
              "mean": 108.7,
              "p95": 108.7,
              "max": 108.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1079,
              "min": 1079,
              "median": 1079,
              "mean": 1079,
              "p95": 1079,
              "max": 1079
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-30",
      "run": "knobs/b70-moe-nvme-off-ncpu40-p1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 62.604934131014176,
      "decode": 15.861662070607508,
      "ttft": null,
      "latency": 96.35,
      "wall": 192.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 192.7,
        "median_decode_tps": 15.862,
        "median_prefill_tps": 62.605,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 31.723324141215016,
            "min": 15.78550405637688,
            "median": 15.861662070607508,
            "mean": 15.861662070607508,
            "p95": 15.937820084838137,
            "max": 15.937820084838137
          },
          "prefill_tps": {
            "n": 2,
            "total": 125.20986826202835,
            "min": 61.19163877447786,
            "median": 62.604934131014176,
            "mean": 62.604934131014176,
            "p95": 64.01822948755049,
            "max": 64.01822948755049
          },
          "elapsed_s": {
            "n": 2,
            "total": 192.7,
            "min": 90.7,
            "median": 96.35,
            "mean": 96.35,
            "p95": 102,
            "max": 102
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2072,
            "min": 905,
            "median": 1036,
            "mean": 1036,
            "p95": 1167,
            "max": 1167
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 90.7,
          "median_decode_tps": 15.786,
          "median_prefill_tps": 61.192,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.78550405637688,
              "min": 15.78550405637688,
              "median": 15.78550405637688,
              "mean": 15.78550405637688,
              "p95": 15.78550405637688,
              "max": 15.78550405637688
            },
            "prefill_tps": {
              "n": 1,
              "total": 61.19163877447786,
              "min": 61.19163877447786,
              "median": 61.19163877447786,
              "mean": 61.19163877447786,
              "p95": 61.19163877447786,
              "max": 61.19163877447786
            },
            "elapsed_s": {
              "n": 1,
              "total": 90.7,
              "min": 90.7,
              "median": 90.7,
              "mean": 90.7,
              "p95": 90.7,
              "max": 90.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 905,
              "min": 905,
              "median": 905,
              "mean": 905,
              "p95": 905,
              "max": 905
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 102,
          "median_decode_tps": 15.938,
          "median_prefill_tps": 64.018,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.937820084838137,
              "min": 15.937820084838137,
              "median": 15.937820084838137,
              "mean": 15.937820084838137,
              "p95": 15.937820084838137,
              "max": 15.937820084838137
            },
            "prefill_tps": {
              "n": 1,
              "total": 64.01822948755049,
              "min": 64.01822948755049,
              "median": 64.01822948755049,
              "mean": 64.01822948755049,
              "p95": 64.01822948755049,
              "max": 64.01822948755049
            },
            "elapsed_s": {
              "n": 1,
              "total": 102,
              "min": 102,
              "median": 102,
              "mean": 102,
              "p95": 102,
              "max": 102
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1167,
              "min": 1167,
              "median": 1167,
              "mean": 1167,
              "p95": 1167,
              "max": 1167
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-31",
      "run": "knobs/b70-moe-nvme-off-ncpu40-p2.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 63.43729906783933,
      "decode": 15.929008680686362,
      "ttft": null,
      "latency": 99.9,
      "wall": 199.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 199.8,
        "median_decode_tps": 15.929,
        "median_prefill_tps": 63.437,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 31.858017361372724,
            "min": 15.89559923632179,
            "median": 15.929008680686362,
            "mean": 15.929008680686362,
            "p95": 15.962418125050933,
            "max": 15.962418125050933
          },
          "prefill_tps": {
            "n": 2,
            "total": 126.87459813567865,
            "min": 62.802846717584586,
            "median": 63.43729906783933,
            "mean": 63.43729906783933,
            "p95": 64.07175141809407,
            "max": 64.07175141809407
          },
          "elapsed_s": {
            "n": 2,
            "total": 199.8,
            "min": 95.3,
            "median": 99.9,
            "mean": 99.9,
            "p95": 104.5,
            "max": 104.5
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2206,
            "min": 997,
            "median": 1103,
            "mean": 1103,
            "p95": 1209,
            "max": 1209
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 95.3,
          "median_decode_tps": 15.896,
          "median_prefill_tps": 62.803,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.89559923632179,
              "min": 15.89559923632179,
              "median": 15.89559923632179,
              "mean": 15.89559923632179,
              "p95": 15.89559923632179,
              "max": 15.89559923632179
            },
            "prefill_tps": {
              "n": 1,
              "total": 62.802846717584586,
              "min": 62.802846717584586,
              "median": 62.802846717584586,
              "mean": 62.802846717584586,
              "p95": 62.802846717584586,
              "max": 62.802846717584586
            },
            "elapsed_s": {
              "n": 1,
              "total": 95.3,
              "min": 95.3,
              "median": 95.3,
              "mean": 95.3,
              "p95": 95.3,
              "max": 95.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 997,
              "min": 997,
              "median": 997,
              "mean": 997,
              "p95": 997,
              "max": 997
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 104.5,
          "median_decode_tps": 15.962,
          "median_prefill_tps": 64.072,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 15.962418125050933,
              "min": 15.962418125050933,
              "median": 15.962418125050933,
              "mean": 15.962418125050933,
              "p95": 15.962418125050933,
              "max": 15.962418125050933
            },
            "prefill_tps": {
              "n": 1,
              "total": 64.07175141809407,
              "min": 64.07175141809407,
              "median": 64.07175141809407,
              "mean": 64.07175141809407,
              "p95": 64.07175141809407,
              "max": 64.07175141809407
            },
            "elapsed_s": {
              "n": 1,
              "total": 104.5,
              "min": 104.5,
              "median": 104.5,
              "mean": 104.5,
              "p95": 104.5,
              "max": 104.5
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1209,
              "min": 1209,
              "median": 1209,
              "mean": 1209,
              "p95": 1209,
              "max": 1209
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-32",
      "run": "knobs/flag-rp1.1-all24.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "`27b-q2-cuda`'s server plus `--repeat-penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.056,
      "structure": 22,
      "strict": null,
      "prefill": 668.022196581089,
      "decode": 41.47599773894104,
      "ttft": null,
      "latency": 32.8,
      "wall": 795.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 144,
        "check_percent": 93.056,
        "wall_s": 795.7,
        "median_decode_tps": 41.476,
        "median_prefill_tps": 668.022,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 991.5900022757619,
            "min": 39.58009262020799,
            "median": 41.47599773894104,
            "mean": 41.31625009482342,
            "p95": 43.16828218594307,
            "max": 43.87555115695733
          },
          "prefill_tps": {
            "n": 24,
            "total": 15923.762967977467,
            "min": 603.8817659001915,
            "median": 668.022196581089,
            "mean": 663.4901236657279,
            "p95": 688.9220881596959,
            "max": 712.8636623054418
          },
          "elapsed_s": {
            "n": 24,
            "total": 795.7,
            "min": 17,
            "median": 32.8,
            "mean": 33.15416666666667,
            "p95": 45,
            "max": 56.4
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 28820,
            "min": 617,
            "median": 1173,
            "mean": 1200.8333333333333,
            "p95": 1758,
            "max": 2152
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 68,
          "check_percent": 98.529,
          "wall_s": 360.2,
          "median_decode_tps": 41.876,
          "median_prefill_tps": 660.714,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 504.49415706296486,
              "min": 40.70106378004923,
              "median": 41.87605018366755,
              "mean": 42.04117975524707,
              "p95": 43.87555115695733,
              "max": 43.87555115695733
            },
            "prefill_tps": {
              "n": 12,
              "total": 7909.657407110352,
              "min": 603.8817659001915,
              "median": 660.7136457025606,
              "mean": 659.138117259196,
              "p95": 712.8636623054418,
              "max": 712.8636623054418
            },
            "elapsed_s": {
              "n": 12,
              "total": 360.2,
              "min": 17,
              "median": 26.049999999999997,
              "mean": 30.016666666666666,
              "p95": 56.4,
              "max": 56.4
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13430,
              "min": 617,
              "median": 972.5,
              "mean": 1119.1666666666667,
              "p95": 2152,
              "max": 2152
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 435.5,
          "median_decode_tps": 40.426,
          "median_prefill_tps": 671.016,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 487.0958452127971,
              "min": 39.58009262020799,
              "median": 40.42627659475282,
              "mean": 40.59132043439976,
              "p95": 41.785436027715804,
              "max": 41.785436027715804
            },
            "prefill_tps": {
              "n": 12,
              "total": 8014.105560867116,
              "min": 621.4310404645125,
              "median": 671.0162674393628,
              "mean": 667.8421300722597,
              "p95": 688.9220881596959,
              "max": 688.9220881596959
            },
            "elapsed_s": {
              "n": 12,
              "total": 435.5,
              "min": 30.8,
              "median": 35.1,
              "mean": 36.291666666666664,
              "p95": 45,
              "max": 45
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15390,
              "min": 1006,
              "median": 1250,
              "mean": 1282.5,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-33",
      "run": "knobs/flag-rp1.1-stream.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 78.125,
      "structure": 4,
      "strict": null,
      "prefill": 22.509594714747163,
      "decode": 39.886560208448266,
      "ttft": null,
      "latency": 30.6,
      "wall": 162.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 25,
        "checks_possible": 32,
        "check_percent": 78.125,
        "wall_s": 162.6,
        "median_decode_tps": 39.887,
        "median_prefill_tps": 22.51,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 200.5527159017592,
            "min": 39.539151958464885,
            "median": 39.886560208448266,
            "mean": 40.11054318035184,
            "p95": 40.89145907408239,
            "max": 40.89145907408239
          },
          "prefill_tps": {
            "n": 5,
            "total": 112.7574418727962,
            "min": 22.298905681203696,
            "median": 22.509594714747163,
            "mean": 22.55148837455924,
            "p95": 22.91515092491278,
            "max": 22.91515092491278
          },
          "elapsed_s": {
            "n": 5,
            "total": 162.6,
            "min": 25.9,
            "median": 30.6,
            "mean": 32.52,
            "p95": 40.3,
            "max": 40.3
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": null,
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 25,
          "checks_possible": 32,
          "check_percent": 78.125,
          "wall_s": 162.6,
          "median_decode_tps": 39.887,
          "median_prefill_tps": 22.51,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 200.5527159017592,
              "min": 39.539151958464885,
              "median": 39.886560208448266,
              "mean": 40.11054318035184,
              "p95": 40.89145907408239,
              "max": 40.89145907408239
            },
            "prefill_tps": {
              "n": 5,
              "total": 112.7574418727962,
              "min": 22.298905681203696,
              "median": 22.509594714747163,
              "mean": 22.55148837455924,
              "p95": 22.91515092491278,
              "max": 22.91515092491278
            },
            "elapsed_s": {
              "n": 5,
              "total": 162.6,
              "min": 25.9,
              "median": 30.6,
              "mean": 32.52,
              "p95": 40.3,
              "max": 40.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": null,
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-34",
      "run": "knobs/flag-rp1.1-topk40-all24.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "as above, plus request `top_k 40`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.056,
      "structure": 22,
      "strict": null,
      "prefill": 671.076560779795,
      "decode": 41.125488040223516,
      "ttft": null,
      "latency": 32.900000000000006,
      "wall": 798.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 144,
        "check_percent": 93.056,
        "wall_s": 798.6,
        "median_decode_tps": 41.125,
        "median_prefill_tps": 671.077,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 988.5394454619561,
            "min": 39.44721707811904,
            "median": 41.125488040223516,
            "mean": 41.189143560914836,
            "p95": 43.020298560872014,
            "max": 43.29199740159312
          },
          "prefill_tps": {
            "n": 24,
            "total": 15968.140137299219,
            "min": 602.245352483978,
            "median": 671.076560779795,
            "mean": 665.3391723874674,
            "p95": 690.6245177311873,
            "max": 692.8920311099537
          },
          "elapsed_s": {
            "n": 24,
            "total": 798.6,
            "min": 17.1,
            "median": 32.900000000000006,
            "mean": 33.275,
            "p95": 45.7,
            "max": 56.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 28820,
            "min": 617,
            "median": 1173,
            "mean": 1200.8333333333333,
            "p95": 1758,
            "max": 2152
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 68,
          "check_percent": 98.529,
          "wall_s": 363,
          "median_decode_tps": 41.722,
          "median_prefill_tps": 653.052,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 501.59624078227506,
              "min": 40.091470606548974,
              "median": 41.7222879772994,
              "mean": 41.799686731856255,
              "p95": 43.29199740159312,
              "max": 43.29199740159312
            },
            "prefill_tps": {
              "n": 12,
              "total": 7874.644220943442,
              "min": 602.245352483978,
              "median": 653.0515950730353,
              "mean": 656.2203517452868,
              "p95": 690.6245177311873,
              "max": 690.6245177311873
            },
            "elapsed_s": {
              "n": 12,
              "total": 363,
              "min": 17.1,
              "median": 26.45,
              "mean": 30.25,
              "p95": 56.6,
              "max": 56.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13430,
              "min": 617,
              "median": 972.5,
              "mean": 1119.1666666666667,
              "p95": 2152,
              "max": 2152
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 435.6,
          "median_decode_tps": 40.442,
          "median_prefill_tps": 675.711,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 486.943204679681,
              "min": 39.44721707811904,
              "median": 40.44209922743684,
              "mean": 40.57860038997342,
              "p95": 41.946629210853445,
              "max": 41.946629210853445
            },
            "prefill_tps": {
              "n": 12,
              "total": 8093.495916355777,
              "min": 647.3506275488991,
              "median": 675.7108168592604,
              "mean": 674.457993029648,
              "p95": 692.8920311099537,
              "max": 692.8920311099537
            },
            "elapsed_s": {
              "n": 12,
              "total": 435.6,
              "min": 30.9,
              "median": 35,
              "mean": 36.3,
              "p95": 45.2,
              "max": 45.2
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15390,
              "min": 1006,
              "median": 1250,
              "mean": 1282.5,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-35",
      "run": "knobs/gemma-4-12b-qat-a770-opencl-rp1.1.jsonl",
      "model": "Gemma 4-12B QAT · Q4_0",
      "model_family": "Gemma 4-12B QAT",
      "quant": "Q4_0",
      "runtime": "llama.cpp SYCL · OpenCL runtime",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 3,
      "expected": 24,
      "success": 100,
      "quality": 26.667,
      "structure": 3,
      "strict": null,
      "prefill": 532.7442724021222,
      "decode": 11.703589487460308,
      "ttft": null,
      "latency": 91.4,
      "wall": 329,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 3,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 3,
        "ok": 3,
        "structure": 3,
        "strict_passes": null,
        "checks_earned": 4,
        "checks_possible": 15,
        "check_percent": 26.667,
        "wall_s": 329,
        "median_decode_tps": 11.704,
        "median_prefill_tps": 532.744,
        "retries": null,
        "truncated": 0,
        "near_quotes": 10,
        "fabricated_quotes": 1,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 3,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 3,
        "cached_prompt_tokens_recorded_cases": 3,
        "elapsed_s_recorded_cases": 3,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 3,
            "total": 34.95264928445398,
            "min": 11.239869552350735,
            "median": 11.703589487460308,
            "mean": 11.650883094817994,
            "p95": 12.009190244642939,
            "max": 12.009190244642939
          },
          "prefill_tps": {
            "n": 3,
            "total": 1557.591461641854,
            "min": 434.54712118646853,
            "median": 532.7442724021222,
            "mean": 519.197153880618,
            "p95": 590.3000680532633,
            "max": 590.3000680532633
          },
          "elapsed_s": {
            "n": 3,
            "total": 329,
            "min": 78.3,
            "median": 91.4,
            "mean": 109.66666666666667,
            "p95": 159.3,
            "max": 159.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 3,
            "total": 7035,
            "min": 1599,
            "median": 2008,
            "mean": 2345,
            "p95": 3428,
            "max": 3428
          },
          "completion_tokens": {
            "n": 3,
            "total": 3636,
            "min": 894,
            "median": 1023,
            "mean": 1212,
            "p95": 1719,
            "max": 1719
          },
          "cached_prompt_tokens": {
            "n": 3,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": null,
          "checks_earned": 4,
          "checks_possible": 15,
          "check_percent": 26.667,
          "wall_s": 329,
          "median_decode_tps": 11.704,
          "median_prefill_tps": 532.744,
          "retries": null,
          "truncated": 0,
          "near_quotes": 10,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 3,
          "elapsed_s_recorded_cases": 3,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 34.95264928445398,
              "min": 11.239869552350735,
              "median": 11.703589487460308,
              "mean": 11.650883094817994,
              "p95": 12.009190244642939,
              "max": 12.009190244642939
            },
            "prefill_tps": {
              "n": 3,
              "total": 1557.591461641854,
              "min": 434.54712118646853,
              "median": 532.7442724021222,
              "mean": 519.197153880618,
              "p95": 590.3000680532633,
              "max": 590.3000680532633
            },
            "elapsed_s": {
              "n": 3,
              "total": 329,
              "min": 78.3,
              "median": 91.4,
              "mean": 109.66666666666667,
              "p95": 159.3,
              "max": 159.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 7035,
              "min": 1599,
              "median": 2008,
              "mean": 2345,
              "p95": 3428,
              "max": 3428
            },
            "completion_tokens": {
              "n": 3,
              "total": 3636,
              "min": 894,
              "median": 1023,
              "mean": 1212,
              "p95": 1719,
              "max": 1719
            },
            "cached_prompt_tokens": {
              "n": 3,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-36",
      "run": "knobs/granite-4.2-8b-a770-opencl-rp1.1.jsonl",
      "model": "Granite 4.2-8B · Q4_K_M",
      "model_family": "Granite 4.2-8B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp SYCL · OpenCL runtime",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 77.273,
      "structure": 1,
      "strict": null,
      "prefill": 600.8833805299489,
      "decode": 19.021198738169282,
      "ttft": null,
      "latency": 136.1,
      "wall": 272.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 1,
        "strict_passes": null,
        "checks_earned": 17,
        "checks_possible": 22,
        "check_percent": 77.273,
        "wall_s": 272.2,
        "median_decode_tps": 19.021,
        "median_prefill_tps": 600.883,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 1,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 38.042397476338564,
            "min": 17.325533358879074,
            "median": 19.021198738169282,
            "mean": 19.021198738169282,
            "p95": 20.71686411745949,
            "max": 20.71686411745949
          },
          "prefill_tps": {
            "n": 2,
            "total": 1201.7667610598978,
            "min": 88.04687209846986,
            "median": 600.8833805299489,
            "mean": 600.8833805299489,
            "p95": 1113.7198889614278,
            "max": 1113.7198889614278
          },
          "elapsed_s": {
            "n": 2,
            "total": 272.2,
            "min": 64.1,
            "median": 136.1,
            "mean": 136.1,
            "p95": 208.1,
            "max": 208.1
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 5518,
            "min": 2071,
            "median": 2759,
            "mean": 2759,
            "p95": 3447,
            "max": 3447
          },
          "completion_tokens": {
            "n": 2,
            "total": 4214,
            "min": 1286,
            "median": 2107,
            "mean": 2107,
            "p95": 2928,
            "max": 2928
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 2,
          "ok": 2,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 17,
          "checks_possible": 22,
          "check_percent": 77.273,
          "wall_s": 272.2,
          "median_decode_tps": 19.021,
          "median_prefill_tps": 600.883,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 2,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 2,
          "cached_prompt_tokens_recorded_cases": 2,
          "elapsed_s_recorded_cases": 2,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 2,
              "total": 38.042397476338564,
              "min": 17.325533358879074,
              "median": 19.021198738169282,
              "mean": 19.021198738169282,
              "p95": 20.71686411745949,
              "max": 20.71686411745949
            },
            "prefill_tps": {
              "n": 2,
              "total": 1201.7667610598978,
              "min": 88.04687209846986,
              "median": 600.8833805299489,
              "mean": 600.8833805299489,
              "p95": 1113.7198889614278,
              "max": 1113.7198889614278
            },
            "elapsed_s": {
              "n": 2,
              "total": 272.2,
              "min": 64.1,
              "median": 136.1,
              "mean": 136.1,
              "p95": 208.1,
              "max": 208.1
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 2,
              "total": 5518,
              "min": 2071,
              "median": 2759,
              "mean": 2759,
              "p95": 3447,
              "max": 3447
            },
            "completion_tokens": {
              "n": 2,
              "total": 4214,
              "min": 1286,
              "median": 2107,
              "mean": 2107,
              "p95": 2928,
              "max": 2928
            },
            "cached_prompt_tokens": {
              "n": 2,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-37",
      "run": "knobs/lms-rock-replicate.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 77.419,
      "structure": 4,
      "strict": null,
      "prefill": null,
      "decode": 46.56108618610364,
      "ttft": 3.77586,
      "latency": 30,
      "wall": 159.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 24,
        "checks_possible": 31,
        "check_percent": 77.419,
        "wall_s": 159.4,
        "median_decode_tps": 46.561,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 233.03290009146727,
            "min": 44.61928803709814,
            "median": 46.56108618610364,
            "mean": 46.60658001829346,
            "p95": 48.25946105767112,
            "max": 48.25946105767112
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 5,
            "total": 159.4,
            "min": 28.1,
            "median": 30,
            "mean": 31.88,
            "p95": 40.4,
            "max": 40.4
          },
          "ttft_s": {
            "n": 5,
            "total": 19.156795,
            "min": 2.5223240000000002,
            "median": 3.77586,
            "mean": 3.831359,
            "p95": 4.936402,
            "max": 4.936402
          },
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6379,
            "min": 1071,
            "median": 1198,
            "mean": 1275.8,
            "p95": 1641,
            "max": 1641
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 24,
          "checks_possible": 31,
          "check_percent": 77.419,
          "wall_s": 159.4,
          "median_decode_tps": 46.561,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 233.03290009146727,
              "min": 44.61928803709814,
              "median": 46.56108618610364,
              "mean": 46.60658001829346,
              "p95": 48.25946105767112,
              "max": 48.25946105767112
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 5,
              "total": 159.4,
              "min": 28.1,
              "median": 30,
              "mean": 31.88,
              "p95": 40.4,
              "max": 40.4
            },
            "ttft_s": {
              "n": 5,
              "total": 19.156795,
              "min": 2.5223240000000002,
              "median": 3.77586,
              "mean": 3.831359,
              "p95": 4.936402,
              "max": 4.936402
            },
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6379,
              "min": 1071,
              "median": 1198,
              "mean": 1275.8,
              "p95": 1641,
              "max": 1641
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-38",
      "run": "knobs/lms-rock-rp1.0.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 30.303,
      "structure": 4,
      "strict": null,
      "prefill": null,
      "decode": 53.955080920477144,
      "ttft": 0.177815,
      "latency": 24.4,
      "wall": 130.5,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 33,
        "check_percent": 30.303,
        "wall_s": 130.5,
        "median_decode_tps": 53.955,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 21,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 269.3261591270574,
            "min": 51.03570381828505,
            "median": 53.955080920477144,
            "mean": 53.865231825411485,
            "p95": 56.00708433068805,
            "max": 56.00708433068805
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 5,
            "total": 130.5,
            "min": 21,
            "median": 24.4,
            "mean": 26.1,
            "p95": 32,
            "max": 32
          },
          "ttft_s": {
            "n": 5,
            "total": 0.889783,
            "min": 0.175071,
            "median": 0.177815,
            "mean": 0.1779566,
            "p95": 0.180461,
            "max": 0.180461
          },
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6820,
            "min": 1140,
            "median": 1301,
            "mean": 1364,
            "p95": 1599,
            "max": 1599
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 10,
          "checks_possible": 33,
          "check_percent": 30.303,
          "wall_s": 130.5,
          "median_decode_tps": 53.955,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 269.3261591270574,
              "min": 51.03570381828505,
              "median": 53.955080920477144,
              "mean": 53.865231825411485,
              "p95": 56.00708433068805,
              "max": 56.00708433068805
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 5,
              "total": 130.5,
              "min": 21,
              "median": 24.4,
              "mean": 26.1,
              "p95": 32,
              "max": 32
            },
            "ttft_s": {
              "n": 5,
              "total": 0.889783,
              "min": 0.175071,
              "median": 0.177815,
              "mean": 0.1779566,
              "p95": 0.180461,
              "max": 0.180461
            },
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6820,
              "min": 1140,
              "median": 1301,
              "mean": 1364,
              "p95": 1599,
              "max": 1599
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-39",
      "run": "knobs/moe-1card-ncpumoe40-rp1.1-with27b.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 1,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 1,
      "strict": null,
      "prefill": 19.03112540560086,
      "decode": 17.157550983803098,
      "ttft": null,
      "latency": 57.6,
      "wall": 57.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 2036,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Explicit single-GPU run artifact name.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": null,
        "checks_earned": 5,
        "checks_possible": 5,
        "check_percent": 100,
        "wall_s": 57.6,
        "median_decode_tps": 17.158,
        "median_prefill_tps": 19.031,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 1,
        "elapsed_s_recorded_cases": 1,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 17.157550983803098,
            "min": 17.157550983803098,
            "median": 17.157550983803098,
            "mean": 17.157550983803098,
            "p95": 17.157550983803098,
            "max": 17.157550983803098
          },
          "prefill_tps": {
            "n": 1,
            "total": 19.03112540560086,
            "min": 19.03112540560086,
            "median": 19.03112540560086,
            "mean": 19.03112540560086,
            "p95": 19.03112540560086,
            "max": 19.03112540560086
          },
          "elapsed_s": {
            "n": 1,
            "total": 57.6,
            "min": 57.6,
            "median": 57.6,
            "mean": 57.6,
            "p95": 57.6,
            "max": 57.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 1,
            "total": 2040,
            "min": 2040,
            "median": 2040,
            "mean": 2040,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 1,
            "total": 979,
            "min": 979,
            "median": 979,
            "mean": 979,
            "p95": 979,
            "max": 979
          },
          "cached_prompt_tokens": {
            "n": 1,
            "total": 2036,
            "min": 2036,
            "median": 2036,
            "mean": 2036,
            "p95": 2036,
            "max": 2036
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 57.6,
          "median_decode_tps": 17.158,
          "median_prefill_tps": 19.031,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 17.157550983803098,
              "min": 17.157550983803098,
              "median": 17.157550983803098,
              "mean": 17.157550983803098,
              "p95": 17.157550983803098,
              "max": 17.157550983803098
            },
            "prefill_tps": {
              "n": 1,
              "total": 19.03112540560086,
              "min": 19.03112540560086,
              "median": 19.03112540560086,
              "mean": 19.03112540560086,
              "p95": 19.03112540560086,
              "max": 19.03112540560086
            },
            "elapsed_s": {
              "n": 1,
              "total": 57.6,
              "min": 57.6,
              "median": 57.6,
              "mean": 57.6,
              "p95": 57.6,
              "max": 57.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 979,
              "min": 979,
              "median": 979,
              "mean": 979,
              "p95": 979,
              "max": 979
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 2036,
              "min": 2036,
              "median": 2036,
              "mean": 2036,
              "p95": 2036,
              "max": 2036
            }
          }
        }
      }
    },
    {
      "id": "run-40",
      "run": "knobs/moe-1card-ncpumoe40-rp1.1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 128.35083918083708,
      "decode": 16.11928318297335,
      "ttft": null,
      "latency": 81,
      "wall": 162,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Explicit single-GPU run artifact name.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 162,
        "median_decode_tps": 16.119,
        "median_prefill_tps": 128.351,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 32.2385663659467,
            "min": 14.838312080761225,
            "median": 16.11928318297335,
            "mean": 16.11928318297335,
            "p95": 17.40025428518548,
            "max": 17.40025428518548
          },
          "prefill_tps": {
            "n": 2,
            "total": 256.70167836167417,
            "min": 126.77513149191212,
            "median": 128.35083918083708,
            "mean": 128.35083918083708,
            "p95": 129.92654686976203,
            "max": 129.92654686976203
          },
          "elapsed_s": {
            "n": 2,
            "total": 162,
            "min": 72.4,
            "median": 81,
            "mean": 81,
            "p95": 89.6,
            "max": 89.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2099,
            "min": 981,
            "median": 1049.5,
            "mean": 1049.5,
            "p95": 1118,
            "max": 1118
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 72.4,
          "median_decode_tps": 17.4,
          "median_prefill_tps": 126.775,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 17.40025428518548,
              "min": 17.40025428518548,
              "median": 17.40025428518548,
              "mean": 17.40025428518548,
              "p95": 17.40025428518548,
              "max": 17.40025428518548
            },
            "prefill_tps": {
              "n": 1,
              "total": 126.77513149191212,
              "min": 126.77513149191212,
              "median": 126.77513149191212,
              "mean": 126.77513149191212,
              "p95": 126.77513149191212,
              "max": 126.77513149191212
            },
            "elapsed_s": {
              "n": 1,
              "total": 72.4,
              "min": 72.4,
              "median": 72.4,
              "mean": 72.4,
              "p95": 72.4,
              "max": 72.4
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 981,
              "min": 981,
              "median": 981,
              "mean": 981,
              "p95": 981,
              "max": 981
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 89.6,
          "median_decode_tps": 14.838,
          "median_prefill_tps": 129.927,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.838312080761225,
              "min": 14.838312080761225,
              "median": 14.838312080761225,
              "mean": 14.838312080761225,
              "p95": 14.838312080761225,
              "max": 14.838312080761225
            },
            "prefill_tps": {
              "n": 1,
              "total": 129.92654686976203,
              "min": 129.92654686976203,
              "median": 129.92654686976203,
              "mean": 129.92654686976203,
              "p95": 129.92654686976203,
              "max": 129.92654686976203
            },
            "elapsed_s": {
              "n": 1,
              "total": 89.6,
              "min": 89.6,
              "median": 89.6,
              "mean": 89.6,
              "p95": 89.6,
              "max": 89.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1118,
              "min": 1118,
              "median": 1118,
              "mean": 1118,
              "p95": 1118,
              "max": 1118
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-41",
      "run": "knobs/moe-1card-ncpumoe48-rp1.1.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 111.26887156357938,
      "decode": 13.056127136915862,
      "ttft": null,
      "latency": 98.5,
      "wall": 197,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Explicit single-GPU run artifact name.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 197,
        "median_decode_tps": 13.056,
        "median_prefill_tps": 111.269,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 2,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 26.112254273831724,
            "min": 11.934790504051053,
            "median": 13.056127136915862,
            "mean": 13.056127136915862,
            "p95": 14.177463769780672,
            "max": 14.177463769780672
          },
          "prefill_tps": {
            "n": 2,
            "total": 222.53774312715876,
            "min": 110.6644690728341,
            "median": 111.26887156357938,
            "mean": 111.26887156357938,
            "p95": 111.87327405432467,
            "max": 111.87327405432467
          },
          "elapsed_s": {
            "n": 2,
            "total": 197,
            "min": 87.6,
            "median": 98.5,
            "mean": 98.5,
            "p95": 109.4,
            "max": 109.4
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 3874,
            "min": 1834,
            "median": 1937,
            "mean": 1937,
            "p95": 2040,
            "max": 2040
          },
          "completion_tokens": {
            "n": 2,
            "total": 2089,
            "min": 981,
            "median": 1044.5,
            "mean": 1044.5,
            "p95": 1108,
            "max": 1108
          },
          "cached_prompt_tokens": {
            "n": 2,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 87.6,
          "median_decode_tps": 14.177,
          "median_prefill_tps": 110.664,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 14.177463769780672,
              "min": 14.177463769780672,
              "median": 14.177463769780672,
              "mean": 14.177463769780672,
              "p95": 14.177463769780672,
              "max": 14.177463769780672
            },
            "prefill_tps": {
              "n": 1,
              "total": 110.6644690728341,
              "min": 110.6644690728341,
              "median": 110.6644690728341,
              "mean": 110.6644690728341,
              "p95": 110.6644690728341,
              "max": 110.6644690728341
            },
            "elapsed_s": {
              "n": 1,
              "total": 87.6,
              "min": 87.6,
              "median": 87.6,
              "mean": 87.6,
              "p95": 87.6,
              "max": 87.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 2040,
              "min": 2040,
              "median": 2040,
              "mean": 2040,
              "p95": 2040,
              "max": 2040
            },
            "completion_tokens": {
              "n": 1,
              "total": 981,
              "min": 981,
              "median": 981,
              "mean": 981,
              "p95": 981,
              "max": 981
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 5,
          "checks_possible": 5,
          "check_percent": 100,
          "wall_s": 109.4,
          "median_decode_tps": 11.935,
          "median_prefill_tps": 111.873,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 1,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 11.934790504051053,
              "min": 11.934790504051053,
              "median": 11.934790504051053,
              "mean": 11.934790504051053,
              "p95": 11.934790504051053,
              "max": 11.934790504051053
            },
            "prefill_tps": {
              "n": 1,
              "total": 111.87327405432467,
              "min": 111.87327405432467,
              "median": 111.87327405432467,
              "mean": 111.87327405432467,
              "p95": 111.87327405432467,
              "max": 111.87327405432467
            },
            "elapsed_s": {
              "n": 1,
              "total": 109.4,
              "min": 109.4,
              "median": 109.4,
              "mean": 109.4,
              "p95": 109.4,
              "max": 109.4
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 1834,
              "min": 1834,
              "median": 1834,
              "mean": 1834,
              "p95": 1834,
              "max": 1834
            },
            "completion_tokens": {
              "n": 1,
              "total": 1108,
              "min": 1108,
              "median": 1108,
              "mean": 1108,
              "p95": 1108,
              "max": 1108
            },
            "cached_prompt_tokens": {
              "n": 1,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-42",
      "run": "knobs/moe-q2-rp1.0-control3.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 3,
      "expected": 24,
      "success": 100,
      "quality": 47.059,
      "structure": 2,
      "strict": null,
      "prefill": 185.6872571555819,
      "decode": 22.931268720188953,
      "ttft": null,
      "latency": 67.8,
      "wall": 226,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 3,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The Flash-Next MoE with the repeat penalty, 2026-09-14",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 3,
        "ok": 3,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 8,
        "checks_possible": 17,
        "check_percent": 47.059,
        "wall_s": 226,
        "median_decode_tps": 22.931,
        "median_prefill_tps": 185.687,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 3,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 3,
        "cached_prompt_tokens_recorded_cases": 3,
        "elapsed_s_recorded_cases": 3,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 3,
            "total": 68.31847265564691,
            "min": 22.35681813615904,
            "median": 22.931268720188953,
            "mean": 22.772824218548973,
            "p95": 23.030385799298926,
            "max": 23.030385799298926
          },
          "prefill_tps": {
            "n": 3,
            "total": 557.8773046086678,
            "min": 184.50198294777053,
            "median": 185.6872571555819,
            "mean": 185.95910153622262,
            "p95": 187.6880645053154,
            "max": 187.6880645053154
          },
          "elapsed_s": {
            "n": 3,
            "total": 226,
            "min": 63.9,
            "median": 67.8,
            "mean": 75.33333333333333,
            "p95": 94.3,
            "max": 94.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 3,
            "total": 10902,
            "min": 2881,
            "median": 3085,
            "mean": 3634,
            "p95": 4936,
            "max": 4936
          },
          "completion_tokens": {
            "n": 3,
            "total": 3794,
            "min": 1084,
            "median": 1200,
            "mean": 1264.6666666666667,
            "p95": 1510,
            "max": 1510
          },
          "cached_prompt_tokens": {
            "n": 3,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 3,
          "ok": 3,
          "structure": 2,
          "strict_passes": null,
          "checks_earned": 8,
          "checks_possible": 17,
          "check_percent": 47.059,
          "wall_s": 226,
          "median_decode_tps": 22.931,
          "median_prefill_tps": 185.687,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 3,
          "elapsed_s_recorded_cases": 3,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 68.31847265564691,
              "min": 22.35681813615904,
              "median": 22.931268720188953,
              "mean": 22.772824218548973,
              "p95": 23.030385799298926,
              "max": 23.030385799298926
            },
            "prefill_tps": {
              "n": 3,
              "total": 557.8773046086678,
              "min": 184.50198294777053,
              "median": 185.6872571555819,
              "mean": 185.95910153622262,
              "p95": 187.6880645053154,
              "max": 187.6880645053154
            },
            "elapsed_s": {
              "n": 3,
              "total": 226,
              "min": 63.9,
              "median": 67.8,
              "mean": 75.33333333333333,
              "p95": 94.3,
              "max": 94.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 10902,
              "min": 2881,
              "median": 3085,
              "mean": 3634,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 3,
              "total": 3794,
              "min": 1084,
              "median": 1200,
              "mean": 1264.6666666666667,
              "p95": 1510,
              "max": 1510
            },
            "cached_prompt_tokens": {
              "n": 3,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-43",
      "run": "knobs/moe-q2-rp1.1-all24.jsonl",
      "model": "Flash-Next MoE Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "`flash-next-q2`'s server, request `repeat_penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 91.538,
      "structure": 22,
      "strict": null,
      "prefill": 185.60743320713954,
      "decode": 18.709627681167422,
      "ttft": null,
      "latency": 83.55,
      "wall": 2064.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 2877,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 2×4060 Ti + Xeon",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 119,
        "checks_possible": 130,
        "check_percent": 91.538,
        "wall_s": 2064.9,
        "median_decode_tps": 18.71,
        "median_prefill_tps": 185.607,
        "retries": null,
        "truncated": 0,
        "near_quotes": 11,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 444.1941828308484,
            "min": 15.132773320288889,
            "median": 18.709627681167422,
            "mean": 18.508090951285347,
            "p95": 20.860902589249704,
            "max": 21.604585641671385
          },
          "prefill_tps": {
            "n": 24,
            "total": 4278.991423461752,
            "min": 21.617298161989222,
            "median": 185.60743320713954,
            "mean": 178.29130931090634,
            "p95": 189.13886348040378,
            "max": 189.38236462179918
          },
          "elapsed_s": {
            "n": 24,
            "total": 2064.9,
            "min": 43.1,
            "median": 83.55,
            "mean": 86.0375,
            "p95": 136.8,
            "max": 172.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 31834,
            "min": 688,
            "median": 1248,
            "mean": 1326.4166666666667,
            "p95": 2127,
            "max": 2521
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 2877,
            "min": 0,
            "median": 0,
            "mean": 119.875,
            "p95": 0,
            "max": 2877
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 65,
          "check_percent": 95.385,
          "wall_s": 852.5,
          "median_decode_tps": 19.532,
          "median_prefill_tps": 183.257,
          "retries": null,
          "truncated": 0,
          "near_quotes": 3,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 230.45634895293446,
              "min": 17.303190515327447,
              "median": 19.53170296123482,
              "mean": 19.20469574607787,
              "p95": 20.860902589249704,
              "max": 20.860902589249704
            },
            "prefill_tps": {
              "n": 12,
              "total": 2196.1920203576183,
              "min": 177.5910150309888,
              "median": 183.25681256692837,
              "mean": 183.0160016964682,
              "p95": 187.46998497596596,
              "max": 187.46998497596596
            },
            "elapsed_s": {
              "n": 12,
              "total": 852.5,
              "min": 43.1,
              "median": 60.7,
              "mean": 71.04166666666667,
              "p95": 127.3,
              "max": 127.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13607,
              "min": 688,
              "median": 990,
              "mean": 1133.9166666666667,
              "p95": 1974,
              "max": 1974
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 57,
          "checks_possible": 65,
          "check_percent": 87.692,
          "wall_s": 1212.4,
          "median_decode_tps": 17.971,
          "median_prefill_tps": 186.97,
          "retries": null,
          "truncated": 0,
          "near_quotes": 8,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 213.73783387791389,
              "min": 15.132773320288889,
              "median": 17.971487864978336,
              "mean": 17.811486156492826,
              "p95": 21.604585641671385,
              "max": 21.604585641671385
            },
            "prefill_tps": {
              "n": 12,
              "total": 2082.7994031041335,
              "min": 21.617298161989222,
              "median": 186.97034952272259,
              "mean": 173.56661692534448,
              "p95": 189.38236462179918,
              "max": 189.38236462179918
            },
            "elapsed_s": {
              "n": 12,
              "total": 1212.4,
              "min": 66.3,
              "median": 90.44999999999999,
              "mean": 101.03333333333333,
              "p95": 172.6,
              "max": 172.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 18227,
              "min": 980,
              "median": 1424.5,
              "mean": 1518.9166666666667,
              "p95": 2521,
              "max": 2521
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 2877,
              "min": 0,
              "median": 0,
              "mean": 239.75,
              "p95": 2877,
              "max": 2877
            }
          }
        }
      }
    },
    {
      "id": "run-44",
      "run": "knobs/moe-q2-rp1.1-probe3.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 3,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": 21.642328065229975,
      "decode": 15.660517037501931,
      "ttft": null,
      "latency": 88.2,
      "wall": 248.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 10890,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 3,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The Flash-Next MoE with the repeat penalty, 2026-09-14",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 3,
        "ok": 3,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 15,
        "checks_possible": 15,
        "check_percent": 100,
        "wall_s": 248.1,
        "median_decode_tps": 15.661,
        "median_prefill_tps": 21.642,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 3,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 3,
        "cached_prompt_tokens_recorded_cases": 3,
        "elapsed_s_recorded_cases": 3,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 3,
            "total": 46.622251540742475,
            "min": 14.497480028039064,
            "median": 15.660517037501931,
            "mean": 15.540750513580825,
            "p95": 16.46425447520148,
            "max": 16.46425447520148
          },
          "prefill_tps": {
            "n": 3,
            "total": 63.83980830326755,
            "min": 20.27174271104151,
            "median": 21.642328065229975,
            "mean": 21.279936101089184,
            "p95": 21.925737526996066,
            "max": 21.925737526996066
          },
          "elapsed_s": {
            "n": 3,
            "total": 248.1,
            "min": 61.3,
            "median": 88.2,
            "mean": 82.7,
            "p95": 98.6,
            "max": 98.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 3,
            "total": 10902,
            "min": 2881,
            "median": 3085,
            "mean": 3634,
            "p95": 4936,
            "max": 4936
          },
          "completion_tokens": {
            "n": 3,
            "total": 3799,
            "min": 1002,
            "median": 1374,
            "mean": 1266.3333333333333,
            "p95": 1423,
            "max": 1423
          },
          "cached_prompt_tokens": {
            "n": 3,
            "total": 10890,
            "min": 2877,
            "median": 3081,
            "mean": 3630,
            "p95": 4932,
            "max": 4932
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 3,
          "ok": 3,
          "structure": 2,
          "strict_passes": null,
          "checks_earned": 15,
          "checks_possible": 15,
          "check_percent": 100,
          "wall_s": 248.1,
          "median_decode_tps": 15.661,
          "median_prefill_tps": 21.642,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 3,
          "elapsed_s_recorded_cases": 3,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 46.622251540742475,
              "min": 14.497480028039064,
              "median": 15.660517037501931,
              "mean": 15.540750513580825,
              "p95": 16.46425447520148,
              "max": 16.46425447520148
            },
            "prefill_tps": {
              "n": 3,
              "total": 63.83980830326755,
              "min": 20.27174271104151,
              "median": 21.642328065229975,
              "mean": 21.279936101089184,
              "p95": 21.925737526996066,
              "max": 21.925737526996066
            },
            "elapsed_s": {
              "n": 3,
              "total": 248.1,
              "min": 61.3,
              "median": 88.2,
              "mean": 82.7,
              "p95": 98.6,
              "max": 98.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 10902,
              "min": 2881,
              "median": 3085,
              "mean": 3634,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 3,
              "total": 3799,
              "min": 1002,
              "median": 1374,
              "mean": 1266.3333333333333,
              "p95": 1423,
              "max": 1423
            },
            "cached_prompt_tokens": {
              "n": 3,
              "total": 10890,
              "min": 2877,
              "median": 3081,
              "mean": 3630,
              "p95": 4932,
              "max": 4932
            }
          }
        }
      }
    },
    {
      "id": "run-45",
      "run": "knobs/prod-acceptance-b10820.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "production `chat-27b-q2`: CUDA b10820, 49k ctx, 2 slots, q8_0 KV, MTP, `--repeat-penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.056,
      "structure": 22,
      "strict": null,
      "prefill": 685.3473781225991,
      "decode": 42.006831724971754,
      "ttft": null,
      "latency": 32.599999999999994,
      "wall": 785.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": 49152,
      "slots": 2,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 144,
        "check_percent": 93.056,
        "wall_s": 785.8,
        "median_decode_tps": 42.007,
        "median_prefill_tps": 685.347,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1003.577372235679,
            "min": 39.81517576244966,
            "median": 42.006831724971754,
            "mean": 41.81572384315329,
            "p95": 43.47408332237555,
            "max": 45.224965632758405
          },
          "prefill_tps": {
            "n": 24,
            "total": 16401.60723050997,
            "min": 635.5594246273733,
            "median": 685.3473781225991,
            "mean": 683.4003012712487,
            "p95": 721.6646397690673,
            "max": 772.9045917273872
          },
          "elapsed_s": {
            "n": 24,
            "total": 785.8,
            "min": 16.8,
            "median": 32.599999999999994,
            "mean": 32.74166666666667,
            "p95": 45.1,
            "max": 55.9
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 28820,
            "min": 617,
            "median": 1173,
            "mean": 1200.8333333333333,
            "p95": 1758,
            "max": 2152
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 68,
          "check_percent": 98.529,
          "wall_s": 352.7,
          "median_decode_tps": 42.66,
          "median_prefill_tps": 689.108,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 513.8887110992708,
              "min": 41.90179457359517,
              "median": 42.66018957495332,
              "mean": 42.82405925827257,
              "p95": 45.224965632758405,
              "max": 45.224965632758405
            },
            "prefill_tps": {
              "n": 12,
              "total": 8304.638243733618,
              "min": 635.5594246273733,
              "median": 689.1081634267066,
              "mean": 692.0531869778015,
              "p95": 772.9045917273872,
              "max": 772.9045917273872
            },
            "elapsed_s": {
              "n": 12,
              "total": 352.7,
              "min": 16.8,
              "median": 25.299999999999997,
              "mean": 29.391666666666666,
              "p95": 55.9,
              "max": 55.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13430,
              "min": 617,
              "median": 972.5,
              "mean": 1119.1666666666667,
              "p95": 2152,
              "max": 2152
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 433.1,
          "median_decode_tps": 40.575,
          "median_prefill_tps": 681.581,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 489.68866113640814,
              "min": 39.81517576244966,
              "median": 40.5750570237494,
              "mean": 40.80738842803401,
              "p95": 42.67400380534174,
              "max": 42.67400380534174
            },
            "prefill_tps": {
              "n": 12,
              "total": 8096.968986776352,
              "min": 641.317203101215,
              "median": 681.5813872152182,
              "mean": 674.747415564696,
              "p95": 691.7811347672581,
              "max": 691.7811347672581
            },
            "elapsed_s": {
              "n": 12,
              "total": 433.1,
              "min": 30.8,
              "median": 34.5,
              "mean": 36.09166666666667,
              "p95": 45.1,
              "max": 45.1
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15390,
              "min": 1006,
              "median": 1250,
              "mean": 1282.5,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-46",
      "run": "knobs/prod-acceptance-b10920.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "as above on CUDA b10920",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 94.483,
      "structure": 22,
      "strict": null,
      "prefill": 689.1537756619912,
      "decode": 42.181510206755796,
      "ttft": null,
      "latency": 33.6,
      "wall": 803.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 137,
        "checks_possible": 145,
        "check_percent": 94.483,
        "wall_s": 803.1,
        "median_decode_tps": 42.182,
        "median_prefill_tps": 689.154,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1009.1913297781709,
            "min": 39.52686492222267,
            "median": 42.181510206755796,
            "mean": 42.04963874075712,
            "p95": 44.70712219125691,
            "max": 45.33654741166923
          },
          "prefill_tps": {
            "n": 24,
            "total": 16615.753208689766,
            "min": 660.6999517076573,
            "median": 689.1537756619912,
            "mean": 692.3230503620737,
            "p95": 738.1404859045164,
            "max": 777.8423705360327
          },
          "elapsed_s": {
            "n": 24,
            "total": 803.1,
            "min": 16.8,
            "median": 33.6,
            "mean": 33.4625,
            "p95": 46.7,
            "max": 55.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 29720,
            "min": 617,
            "median": 1191,
            "mean": 1238.3333333333333,
            "p95": 1780,
            "max": 2159
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 68,
          "checks_possible": 68,
          "check_percent": 100,
          "wall_s": 359.09999999999997,
          "median_decode_tps": 43.067,
          "median_prefill_tps": 698.71,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 518.6929963748111,
              "min": 41.91071493692437,
              "median": 43.06726320430343,
              "mean": 43.22441636456759,
              "p95": 45.33654741166923,
              "max": 45.33654741166923
            },
            "prefill_tps": {
              "n": 12,
              "total": 8494.209144310107,
              "min": 660.6999517076573,
              "median": 698.7097970208159,
              "mean": 707.8507620258423,
              "p95": 777.8423705360327,
              "max": 777.8423705360327
            },
            "elapsed_s": {
              "n": 12,
              "total": 359.09999999999997,
              "min": 16.8,
              "median": 25.35,
              "mean": 29.925,
              "p95": 55.3,
              "max": 55.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13878,
              "min": 617,
              "median": 957.5,
              "mean": 1156.5,
              "p95": 2159,
              "max": 2159
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 77,
          "check_percent": 89.61,
          "wall_s": 444,
          "median_decode_tps": 40.46,
          "median_prefill_tps": 673.849,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 490.4983334033599,
              "min": 39.52686492222267,
              "median": 40.460469566393584,
              "mean": 40.874861116946654,
              "p95": 42.761791746427136,
              "max": 42.761791746427136
            },
            "prefill_tps": {
              "n": 12,
              "total": 8121.54406437966,
              "min": 664.1070284187902,
              "median": 673.8485854880337,
              "mean": 676.7953386983049,
              "p95": 693.2975608457484,
              "max": 693.2975608457484
            },
            "elapsed_s": {
              "n": 12,
              "total": 444,
              "min": 30.3,
              "median": 36.05,
              "mean": 37,
              "p95": 46.7,
              "max": 46.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15842,
              "min": 1039,
              "median": 1329.5,
              "mean": 1320.1666666666667,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-47",
      "run": "knobs/q2mtp-baseline.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 30.303,
      "structure": 4,
      "strict": null,
      "prefill": 22.784103530966444,
      "decode": 48.49949938935565,
      "ttft": null,
      "latency": 31.6,
      "wall": 151.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 14422,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 33,
        "check_percent": 30.303,
        "wall_s": 151.3,
        "median_decode_tps": 48.499,
        "median_prefill_tps": 22.784,
        "retries": null,
        "truncated": 0,
        "near_quotes": 21,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 5,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 240.0052212403969,
            "min": 45.847769769903536,
            "median": 48.49949938935565,
            "mean": 48.00104424807938,
            "p95": 48.93969909497536,
            "max": 48.93969909497536
          },
          "prefill_tps": {
            "n": 5,
            "total": 114.06059760105381,
            "min": 22.354861064538486,
            "median": 22.784103530966444,
            "mean": 22.81211952021076,
            "p95": 23.35643674201064,
            "max": 23.35643674201064
          },
          "elapsed_s": {
            "n": 5,
            "total": 151.3,
            "min": 23.6,
            "median": 31.6,
            "mean": 30.26,
            "p95": 35.6,
            "max": 35.6
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 7113,
            "min": 1130,
            "median": 1490,
            "mean": 1422.6,
            "p95": 1608,
            "max": 1608
          },
          "cached_prompt_tokens": {
            "n": 5,
            "total": 14422,
            "min": 1830,
            "median": 2846,
            "mean": 2884.4,
            "p95": 3805,
            "max": 3805
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 10,
          "checks_possible": 33,
          "check_percent": 30.303,
          "wall_s": 151.3,
          "median_decode_tps": 48.499,
          "median_prefill_tps": 22.784,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 5,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 240.0052212403969,
              "min": 45.847769769903536,
              "median": 48.49949938935565,
              "mean": 48.00104424807938,
              "p95": 48.93969909497536,
              "max": 48.93969909497536
            },
            "prefill_tps": {
              "n": 5,
              "total": 114.06059760105381,
              "min": 22.354861064538486,
              "median": 22.784103530966444,
              "mean": 22.81211952021076,
              "p95": 23.35643674201064,
              "max": 23.35643674201064
            },
            "elapsed_s": {
              "n": 5,
              "total": 151.3,
              "min": 23.6,
              "median": 31.6,
              "mean": 30.26,
              "p95": 35.6,
              "max": 35.6
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 7113,
              "min": 1130,
              "median": 1490,
              "mean": 1422.6,
              "p95": 1608,
              "max": 1608
            },
            "cached_prompt_tokens": {
              "n": 5,
              "total": 14422,
              "min": 1830,
              "median": 2846,
              "mean": 2884.4,
              "p95": 3805,
              "max": 3805
            }
          }
        }
      }
    },
    {
      "id": "run-48",
      "run": "knobs/q2mtp-rp1.1-all24.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "as `27b-q2-cuda`, request `repeat_penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.056,
      "structure": 22,
      "strict": null,
      "prefill": 658.5483955639888,
      "decode": 40.72466460633916,
      "ttft": null,
      "latency": 32.95,
      "wall": 797.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 3321,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 144,
        "check_percent": 93.056,
        "wall_s": 797.7,
        "median_decode_tps": 40.725,
        "median_prefill_tps": 658.548,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 984.0798875139806,
            "min": 39.44702690781072,
            "median": 40.72466460633916,
            "mean": 41.00332864641585,
            "p95": 42.70430755233323,
            "max": 43.63786177855059
          },
          "prefill_tps": {
            "n": 24,
            "total": 15147.690504183362,
            "min": 21.88938200802246,
            "median": 658.5483955639888,
            "mean": 631.1537710076401,
            "p95": 684.5905127868712,
            "max": 685.1103171844419
          },
          "elapsed_s": {
            "n": 24,
            "total": 797.7,
            "min": 17.1,
            "median": 32.95,
            "mean": 33.2375,
            "p95": 45.2,
            "max": 57.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 28820,
            "min": 617,
            "median": 1173,
            "mean": 1200.8333333333333,
            "p95": 1758,
            "max": 2152
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 3321,
            "min": 0,
            "median": 0,
            "mean": 138.375,
            "p95": 0,
            "max": 3321
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 68,
          "check_percent": 98.529,
          "wall_s": 361.8,
          "median_decode_tps": 41.365,
          "median_prefill_tps": 640.92,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 497.4226930384605,
              "min": 39.44702690781072,
              "median": 41.36511031510598,
              "mean": 41.451891086538375,
              "p95": 43.63786177855059,
              "max": 43.63786177855059
            },
            "prefill_tps": {
              "n": 12,
              "total": 7076.071361470354,
              "min": 21.88938200802246,
              "median": 640.9200200056551,
              "mean": 589.6726134558628,
              "p95": 678.3461557074368,
              "max": 678.3461557074368
            },
            "elapsed_s": {
              "n": 12,
              "total": 361.8,
              "min": 17.1,
              "median": 26.6,
              "mean": 30.15,
              "p95": 57.3,
              "max": 57.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13430,
              "min": 617,
              "median": 972.5,
              "mean": 1119.1666666666667,
              "p95": 2152,
              "max": 2152
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 3321,
              "min": 0,
              "median": 0,
              "mean": 276.75,
              "p95": 3321,
              "max": 3321
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 435.9,
          "median_decode_tps": 40.338,
          "median_prefill_tps": 677.603,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 486.65719447552,
              "min": 39.78037752924861,
              "median": 40.33767102739613,
              "mean": 40.55476620629334,
              "p95": 42.36285662498898,
              "max": 42.36285662498898
            },
            "prefill_tps": {
              "n": 12,
              "total": 8071.619142713008,
              "min": 646.5680597493756,
              "median": 677.6034810442554,
              "mean": 672.6349285594173,
              "p95": 685.1103171844419,
              "max": 685.1103171844419
            },
            "elapsed_s": {
              "n": 12,
              "total": 435.9,
              "min": 30.9,
              "median": 34.85,
              "mean": 36.325,
              "p95": 45.2,
              "max": 45.2
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15390,
              "min": 1006,
              "median": 1250,
              "mean": 1282.5,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-49",
      "run": "knobs/q2mtp-rp1.1-topk40.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 78.125,
      "structure": 4,
      "strict": null,
      "prefill": 22.216544660808907,
      "decode": 40.93443248869375,
      "ttft": null,
      "latency": 30.4,
      "wall": 160.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 14422,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 25,
        "checks_possible": 32,
        "check_percent": 78.125,
        "wall_s": 160.4,
        "median_decode_tps": 40.934,
        "median_prefill_tps": 22.217,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 5,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 203.88694492271938,
            "min": 40.056001624061246,
            "median": 40.93443248869375,
            "mean": 40.777388984543876,
            "p95": 41.47170050020275,
            "max": 41.47170050020275
          },
          "prefill_tps": {
            "n": 5,
            "total": 111.17798854406662,
            "min": 21.93595797070453,
            "median": 22.216544660808907,
            "mean": 22.235597708813327,
            "p95": 22.706114756703982,
            "max": 22.706114756703982
          },
          "elapsed_s": {
            "n": 5,
            "total": 160.4,
            "min": 25.2,
            "median": 30.4,
            "mean": 32.08,
            "p95": 40.2,
            "max": 40.2
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6420,
            "min": 1006,
            "median": 1231,
            "mean": 1284,
            "p95": 1587,
            "max": 1587
          },
          "cached_prompt_tokens": {
            "n": 5,
            "total": 14422,
            "min": 1830,
            "median": 2846,
            "mean": 2884.4,
            "p95": 3805,
            "max": 3805
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 25,
          "checks_possible": 32,
          "check_percent": 78.125,
          "wall_s": 160.4,
          "median_decode_tps": 40.934,
          "median_prefill_tps": 22.217,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 5,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 203.88694492271938,
              "min": 40.056001624061246,
              "median": 40.93443248869375,
              "mean": 40.777388984543876,
              "p95": 41.47170050020275,
              "max": 41.47170050020275
            },
            "prefill_tps": {
              "n": 5,
              "total": 111.17798854406662,
              "min": 21.93595797070453,
              "median": 22.216544660808907,
              "mean": 22.235597708813327,
              "p95": 22.706114756703982,
              "max": 22.706114756703982
            },
            "elapsed_s": {
              "n": 5,
              "total": 160.4,
              "min": 25.2,
              "median": 30.4,
              "mean": 32.08,
              "p95": 40.2,
              "max": 40.2
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6420,
              "min": 1006,
              "median": 1231,
              "mean": 1284,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 5,
              "total": 14422,
              "min": 1830,
              "median": 2846,
              "mean": 2884.4,
              "p95": 3805,
              "max": 3805
            }
          }
        }
      }
    },
    {
      "id": "run-50",
      "run": "knobs/q2mtp-rp1.1.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 78.125,
      "structure": 4,
      "strict": null,
      "prefill": 22.246322404827453,
      "decode": 39.559563870025826,
      "ttft": null,
      "latency": 30.9,
      "wall": 165.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 14422,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 25,
        "checks_possible": 32,
        "check_percent": 78.125,
        "wall_s": 165.2,
        "median_decode_tps": 39.56,
        "median_prefill_tps": 22.246,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 5,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 197.81181741206973,
            "min": 38.48806453339683,
            "median": 39.559563870025826,
            "mean": 39.56236348241394,
            "p95": 40.611171050191345,
            "max": 40.611171050191345
          },
          "prefill_tps": {
            "n": 5,
            "total": 111.18537927352548,
            "min": 21.87142880576531,
            "median": 22.246322404827453,
            "mean": 22.237075854705097,
            "p95": 22.603850565943908,
            "max": 22.603850565943908
          },
          "elapsed_s": {
            "n": 5,
            "total": 165.2,
            "min": 25.8,
            "median": 30.9,
            "mean": 33.04,
            "p95": 40.7,
            "max": 40.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6420,
            "min": 1006,
            "median": 1231,
            "mean": 1284,
            "p95": 1587,
            "max": 1587
          },
          "cached_prompt_tokens": {
            "n": 5,
            "total": 14422,
            "min": 1830,
            "median": 2846,
            "mean": 2884.4,
            "p95": 3805,
            "max": 3805
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 25,
          "checks_possible": 32,
          "check_percent": 78.125,
          "wall_s": 165.2,
          "median_decode_tps": 39.56,
          "median_prefill_tps": 22.246,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 5,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 197.81181741206973,
              "min": 38.48806453339683,
              "median": 39.559563870025826,
              "mean": 39.56236348241394,
              "p95": 40.611171050191345,
              "max": 40.611171050191345
            },
            "prefill_tps": {
              "n": 5,
              "total": 111.18537927352548,
              "min": 21.87142880576531,
              "median": 22.246322404827453,
              "mean": 22.237075854705097,
              "p95": 22.603850565943908,
              "max": 22.603850565943908
            },
            "elapsed_s": {
              "n": 5,
              "total": 165.2,
              "min": 25.8,
              "median": 30.9,
              "mean": 33.04,
              "p95": 40.7,
              "max": 40.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6420,
              "min": 1006,
              "median": 1231,
              "mean": 1284,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 5,
              "total": 14422,
              "min": 1830,
              "median": 2846,
              "mean": 2884.4,
              "p95": 3805,
              "max": 3805
            }
          }
        }
      }
    },
    {
      "id": "run-51",
      "run": "knobs/q2mtp-topk40.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 5,
      "expected": 24,
      "success": 100,
      "quality": 30.303,
      "structure": 4,
      "strict": null,
      "prefill": 22.835090683853878,
      "decode": 46.01945135793323,
      "ttft": null,
      "latency": 33.5,
      "wall": 158.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 14422,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived REPORT.md, section: The cause: LM Studio's default repeat penalty",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 10,
        "checks_possible": 33,
        "check_percent": 30.303,
        "wall_s": 158.7,
        "median_decode_tps": 46.019,
        "median_prefill_tps": 22.835,
        "retries": null,
        "truncated": 0,
        "near_quotes": 21,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 5,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 228.57189474918943,
            "min": 44.351142373100735,
            "median": 46.01945135793323,
            "mean": 45.71437894983789,
            "p95": 46.936478107853794,
            "max": 46.936478107853794
          },
          "prefill_tps": {
            "n": 5,
            "total": 114.27435743370238,
            "min": 22.61177281951848,
            "median": 22.835090683853878,
            "mean": 22.854871486740475,
            "p95": 23.141452126120914,
            "max": 23.141452126120914
          },
          "elapsed_s": {
            "n": 5,
            "total": 158.7,
            "min": 24.6,
            "median": 33.5,
            "mean": 31.74,
            "p95": 36.8,
            "max": 36.8
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 7113,
            "min": 1130,
            "median": 1490,
            "mean": 1422.6,
            "p95": 1608,
            "max": 1608
          },
          "cached_prompt_tokens": {
            "n": 5,
            "total": 14422,
            "min": 1830,
            "median": 2846,
            "mean": 2884.4,
            "p95": 3805,
            "max": 3805
          }
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 10,
          "checks_possible": 33,
          "check_percent": 30.303,
          "wall_s": 158.7,
          "median_decode_tps": 46.019,
          "median_prefill_tps": 22.835,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 5,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 228.57189474918943,
              "min": 44.351142373100735,
              "median": 46.01945135793323,
              "mean": 45.71437894983789,
              "p95": 46.936478107853794,
              "max": 46.936478107853794
            },
            "prefill_tps": {
              "n": 5,
              "total": 114.27435743370238,
              "min": 22.61177281951848,
              "median": 22.835090683853878,
              "mean": 22.854871486740475,
              "p95": 23.141452126120914,
              "max": 23.141452126120914
            },
            "elapsed_s": {
              "n": 5,
              "total": 158.7,
              "min": 24.6,
              "median": 33.5,
              "mean": 31.74,
              "p95": 36.8,
              "max": 36.8
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 7113,
              "min": 1130,
              "median": 1490,
              "mean": 1422.6,
              "p95": 1608,
              "max": 1608
            },
            "cached_prompt_tokens": {
              "n": 5,
              "total": 14422,
              "min": 1830,
              "median": 2846,
              "mean": 2884.4,
              "p95": 3805,
              "max": 3805
            }
          }
        }
      }
    },
    {
      "id": "run-52",
      "run": "outputs/27b-q2-b10920.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "SYCL b10920, q8_0 KV, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 80.537,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 14.2,
      "ttft": null,
      "latency": 94.25,
      "wall": 2300.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 120,
        "checks_possible": 149,
        "check_percent": 80.537,
        "wall_s": 2300.7,
        "median_decode_tps": 14.2,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 24,
        "fabricated_quotes": 5,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 341,
            "min": 13.8,
            "median": 14.2,
            "mean": 14.208333333333334,
            "p95": 14.6,
            "max": 14.8
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 2300.7,
            "min": 48.6,
            "median": 94.25,
            "mean": 95.8625,
            "p95": 136.9,
            "max": 154.1
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 30539,
            "min": 641,
            "median": 1274.5,
            "mean": 1272.4583333333333,
            "p95": 1824,
            "max": 2077
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 71,
          "check_percent": 97.183,
          "wall_s": 1065.6,
          "median_decode_tps": 14.3,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 2,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 171.6,
              "min": 13.8,
              "median": 14.3,
              "mean": 14.3,
              "p95": 14.6,
              "max": 14.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1065.6,
              "min": 48.6,
              "median": 74.6,
              "mean": 88.8,
              "p95": 154.1,
              "max": 154.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14321,
              "min": 641,
              "median": 1006.5,
              "mean": 1193.4166666666667,
              "p95": 2077,
              "max": 2077
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 51,
          "checks_possible": 78,
          "check_percent": 65.385,
          "wall_s": 1235.1,
          "median_decode_tps": 14.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 22,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 169.4,
              "min": 13.8,
              "median": 14.1,
              "mean": 14.116666666666667,
              "p95": 14.8,
              "max": 14.8
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1235.1,
              "min": 84.5,
              "median": 101.75,
              "mean": 102.925,
              "p95": 136.9,
              "max": 136.9
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 16218,
              "min": 1133,
              "median": 1296.5,
              "mean": 1351.5,
              "p95": 1763,
              "max": 1763
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-53",
      "run": "outputs/27b-q2-b10964-rp1.1-pinned.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "`chat-lab`'s arguments with the Q2 GGUF on SYCL b10964, plus `--repeat-penalty 1.1`, same pins",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 95.035,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 14.5,
      "ttft": null,
      "latency": 91.7,
      "wall": 2173,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 141,
        "check_percent": 95.035,
        "wall_s": 2173,
        "median_decode_tps": 14.5,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 347.5,
            "min": 13.9,
            "median": 14.5,
            "mean": 14.479166666666666,
            "p95": 15.1,
            "max": 15.3
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 2173,
            "min": 46.1,
            "median": 91.7,
            "mean": 90.54166666666667,
            "p95": 118.9,
            "max": 151.7
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29531,
            "min": 617,
            "median": 1261.5,
            "mean": 1230.4583333333333,
            "p95": 1645,
            "max": 2153
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 67,
          "check_percent": 100,
          "wall_s": 986.1,
          "median_decode_tps": 14.6,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 176.4,
              "min": 14.2,
              "median": 14.6,
              "mean": 14.7,
              "p95": 15.3,
              "max": 15.3
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 986.1,
              "min": 46.1,
              "median": 72.94999999999999,
              "mean": 82.175,
              "p95": 151.7,
              "max": 151.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13678,
              "min": 617,
              "median": 980.5,
              "mean": 1139.8333333333333,
              "p95": 2153,
              "max": 2153
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 74,
          "check_percent": 90.541,
          "wall_s": 1186.9,
          "median_decode_tps": 14.25,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 171.1,
              "min": 13.9,
              "median": 14.25,
              "mean": 14.258333333333333,
              "p95": 14.7,
              "max": 14.7
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1186.9,
              "min": 73.2,
              "median": 100.7,
              "mean": 98.90833333333333,
              "p95": 118.9,
              "max": 118.9
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15853,
              "min": 957,
              "median": 1395,
              "mean": 1321.0833333333333,
              "p95": 1586,
              "max": 1586
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-54",
      "run": "outputs/27b-q2-cuda.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "CUDA b10820, q8_0 KV, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 81.208,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 47.8,
      "ttft": null,
      "latency": 28.4,
      "wall": 717.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 121,
        "checks_possible": 149,
        "check_percent": 81.208,
        "wall_s": 717.4,
        "median_decode_tps": 47.8,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 24,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1145.4,
            "min": 44.5,
            "median": 47.8,
            "mean": 47.725,
            "p95": 50.4,
            "max": 51.1
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 717.4,
            "min": 15.5,
            "median": 28.4,
            "mean": 29.891666666666666,
            "p95": 39.9,
            "max": 48.7
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29636,
            "min": 641,
            "median": 1206.5,
            "mean": 1234.8333333333333,
            "p95": 1804,
            "max": 2084
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 71,
          "check_percent": 97.183,
          "wall_s": 326.7,
          "median_decode_tps": 49.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 587.6,
              "min": 47.7,
              "median": 49.1,
              "mean": 48.96666666666667,
              "p95": 51.1,
              "max": 51.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 326.7,
              "min": 15.5,
              "median": 23.75,
              "mean": 27.225,
              "p95": 48.7,
              "max": 48.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14078,
              "min": 641,
              "median": 1008,
              "mean": 1173.1666666666667,
              "p95": 2084,
              "max": 2084
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 52,
          "checks_possible": 78,
          "check_percent": 66.667,
          "wall_s": 390.7,
          "median_decode_tps": 46.45,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 23,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 557.8,
              "min": 44.5,
              "median": 46.45,
              "mean": 46.483333333333334,
              "p95": 48.1,
              "max": 48.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 390.7,
              "min": 26.7,
              "median": 32.5,
              "mean": 32.55833333333333,
              "p95": 38.4,
              "max": 38.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15558,
              "min": 1050,
              "median": 1294,
              "mean": 1296.5,
              "p95": 1578,
              "max": 1578
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-55",
      "run": "outputs/27b-q2-f16kv.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "as `27b-q2-cuda`, f16 KV",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 82,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 50.05,
      "ttft": null,
      "latency": 29.15,
      "wall": 713,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": "f16",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 123,
        "checks_possible": 150,
        "check_percent": 82,
        "wall_s": 713,
        "median_decode_tps": 50.05,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 24,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1191.4,
            "min": 47.4,
            "median": 50.05,
            "mean": 49.641666666666666,
            "p95": 51.3,
            "max": 51.5
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 713,
            "min": 14.8,
            "median": 29.15,
            "mean": 29.708333333333332,
            "p95": 47.5,
            "max": 53.4
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 30835,
            "min": 641,
            "median": 1222,
            "mean": 1284.7916666666667,
            "p95": 2114,
            "max": 2304
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 71,
          "checks_possible": 72,
          "check_percent": 98.611,
          "wall_s": 325.2,
          "median_decode_tps": 50.25,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 603.3,
              "min": 47.5,
              "median": 50.25,
              "mean": 50.275,
              "p95": 51.5,
              "max": 51.5
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 325.2,
              "min": 14.8,
              "median": 24.85,
              "mean": 27.1,
              "p95": 47.5,
              "max": 47.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14408,
              "min": 641,
              "median": 1092.5,
              "mean": 1200.6666666666667,
              "p95": 2114,
              "max": 2114
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 52,
          "checks_possible": 78,
          "check_percent": 66.667,
          "wall_s": 387.8,
          "median_decode_tps": 49.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 23,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 588.1,
              "min": 47.4,
              "median": 49.1,
              "mean": 49.00833333333333,
              "p95": 50.3,
              "max": 50.3
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 387.8,
              "min": 25.6,
              "median": 31.1,
              "mean": 32.31666666666666,
              "p95": 53.4,
              "max": 53.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 16427,
              "min": 1133,
              "median": 1298,
              "mean": 1368.9166666666667,
              "p95": 2304,
              "max": 2304
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-56",
      "run": "outputs/27b-q2-lab system-lms.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio 2.33.0 Vulkan, headless, 32k, 1 slot, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 95.139,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 3.1,
      "ttft": null,
      "latency": 409.65,
      "wall": 10459.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 32768,
      "slots": 1,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 137,
        "checks_possible": 144,
        "check_percent": 95.139,
        "wall_s": 10459.1,
        "median_decode_tps": 3.1,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 73.6,
            "min": 2.9,
            "median": 3.1,
            "mean": 3.066666666666667,
            "p95": 3.3,
            "max": 3.3
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 10459.1,
            "min": 219.7,
            "median": 409.65,
            "mean": 435.79583333333335,
            "p95": 684.3,
            "max": 722.3
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29258,
            "min": 630,
            "median": 1156.5,
            "mean": 1219.0833333333333,
            "p95": 1844,
            "max": 2158
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 69,
          "check_percent": 100,
          "wall_s": 4703.1,
          "median_decode_tps": 3.15,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 38,
              "min": 3,
              "median": 3.1500000000000004,
              "mean": 3.1666666666666665,
              "p95": 3.3,
              "max": 3.3
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 4703.1,
              "min": 219.7,
              "median": 349.75,
              "mean": 391.925,
              "p95": 722.3,
              "max": 722.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13774,
              "min": 630,
              "median": 1016.5,
              "mean": 1147.8333333333333,
              "p95": 2158,
              "max": 2158
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 68,
          "checks_possible": 75,
          "check_percent": 90.667,
          "wall_s": 5756,
          "median_decode_tps": 2.95,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 35.6,
              "min": 2.9,
              "median": 2.95,
              "mean": 2.966666666666667,
              "p95": 3.1,
              "max": 3.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 5756,
              "min": 361.8,
              "median": 479.79999999999995,
              "mean": 479.6666666666667,
              "p95": 684.3,
              "max": 684.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15484,
              "min": 957,
              "median": 1223.5,
              "mean": 1290.3333333333333,
              "p95": 1844,
              "max": 1844
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-57",
      "run": "outputs/27b-q2-newcuda.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "CUDA b10920, q8_0 KV, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 81.879,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 49.349999999999994,
      "ttft": null,
      "latency": 29.15,
      "wall": 695.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 122,
        "checks_possible": 149,
        "check_percent": 81.879,
        "wall_s": 695.9,
        "median_decode_tps": 49.35,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 23,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1175.9,
            "min": 45.1,
            "median": 49.349999999999994,
            "mean": 48.99583333333333,
            "p95": 50.5,
            "max": 50.6
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 695.9,
            "min": 15,
            "median": 29.15,
            "mean": 28.995833333333334,
            "p95": 40.8,
            "max": 47.6
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29631,
            "min": 641,
            "median": 1220,
            "mean": 1234.625,
            "p95": 1834,
            "max": 2083
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 71,
          "check_percent": 97.183,
          "wall_s": 320.6,
          "median_decode_tps": 50,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 2,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 597.9,
              "min": 48.1,
              "median": 50,
              "mean": 49.825,
              "p95": 50.6,
              "max": 50.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 320.6,
              "min": 15,
              "median": 22.9,
              "mean": 26.716666666666665,
              "p95": 47.6,
              "max": 47.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14060,
              "min": 641,
              "median": 1005.5,
              "mean": 1171.6666666666667,
              "p95": 2083,
              "max": 2083
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 53,
          "checks_possible": 78,
          "check_percent": 67.949,
          "wall_s": 375.3,
          "median_decode_tps": 48.25,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 578,
              "min": 45.1,
              "median": 48.25,
              "mean": 48.166666666666664,
              "p95": 49.8,
              "max": 49.8
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 375.3,
              "min": 26,
              "median": 31.200000000000003,
              "mean": 31.275,
              "p95": 37.3,
              "max": 37.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15571,
              "min": 1140,
              "median": 1282,
              "mean": 1297.5833333333333,
              "p95": 1483,
              "max": 1483
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-58",
      "run": "outputs/27b-q2-nofa.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "f16 KV, flash attention off, no speculation",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 81.879,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 24.5,
      "ttft": null,
      "latency": 52.75,
      "wall": 1289.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": "f16",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 122,
        "checks_possible": 149,
        "check_percent": 81.879,
        "wall_s": 1289.1,
        "median_decode_tps": 24.5,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 23,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 585.6,
            "min": 23.8,
            "median": 24.5,
            "mean": 24.400000000000002,
            "p95": 24.7,
            "max": 24.8
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1289.1,
            "min": 28.7,
            "median": 52.75,
            "mean": 53.7125,
            "p95": 74.9,
            "max": 93.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29293,
            "min": 661,
            "median": 1193.5,
            "mean": 1220.5416666666667,
            "p95": 1804,
            "max": 2084
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 71,
          "check_percent": 97.183,
          "wall_s": 613.3,
          "median_decode_tps": 24.6,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 2,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 294.1,
              "min": 23.8,
              "median": 24.6,
              "mean": 24.508333333333333,
              "p95": 24.8,
              "max": 24.8
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 613.3,
              "min": 28.7,
              "median": 46,
              "mean": 51.108333333333334,
              "p95": 93.5,
              "max": 93.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14136,
              "min": 661,
              "median": 1064,
              "mean": 1178,
              "p95": 2084,
              "max": 2084
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 53,
          "checks_possible": 78,
          "check_percent": 67.949,
          "wall_s": 675.8,
          "median_decode_tps": 24.3,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 291.5,
              "min": 23.8,
              "median": 24.3,
              "mean": 24.291666666666668,
              "p95": 24.6,
              "max": 24.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 675.8,
              "min": 50.6,
              "median": 55.8,
              "mean": 56.31666666666667,
              "p95": 66.4,
              "max": 66.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15157,
              "min": 1145,
              "median": 1226,
              "mean": 1263.0833333333333,
              "p95": 1535,
              "max": 1535
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-59",
      "run": "outputs/27b-q2-lab system-lms.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio 2.33.0 CUDA, 32k, 1 slot, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.151,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 48.7,
      "ttft": null,
      "latency": 28.75,
      "wall": 690.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 32768,
      "slots": 1,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 136,
        "checks_possible": 146,
        "check_percent": 93.151,
        "wall_s": 690.2,
        "median_decode_tps": 48.7,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1162.6,
            "min": 45,
            "median": 48.7,
            "mean": 48.44166666666667,
            "p95": 51.1,
            "max": 52.1
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 690.2,
            "min": 14.6,
            "median": 28.75,
            "mean": 28.758333333333333,
            "p95": 40.1,
            "max": 48.2
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29055,
            "min": 617,
            "median": 1187,
            "mean": 1210.625,
            "p95": 1778,
            "max": 2159
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 70,
          "check_percent": 98.571,
          "wall_s": 317.6,
          "median_decode_tps": 49.15,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 593.4,
              "min": 46.6,
              "median": 49.150000000000006,
              "mean": 49.45,
              "p95": 52.1,
              "max": 52.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 317.6,
              "min": 14.6,
              "median": 23.85,
              "mean": 26.466666666666665,
              "p95": 48.2,
              "max": 48.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13847,
              "min": 617,
              "median": 1036,
              "mean": 1153.9166666666667,
              "p95": 2159,
              "max": 2159
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 372.6,
          "median_decode_tps": 47.05,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 569.2,
              "min": 45,
              "median": 47.05,
              "mean": 47.43333333333334,
              "p95": 49.9,
              "max": 49.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 372.6,
              "min": 24.3,
              "median": 31.200000000000003,
              "mean": 31.05,
              "p95": 40.1,
              "max": 40.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15208,
              "min": 1058,
              "median": 1247,
              "mean": 1267.3333333333333,
              "p95": 1641,
              "max": 1641
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-60",
      "run": "outputs/27b-q2-lab system-r2.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "replicate of `27b-q2-lab system`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.617,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 177.7,
      "ttft": null,
      "latency": 7.949999999999999,
      "wall": 189,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 5090 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core Ultra · RTX 5090",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 132,
        "checks_possible": 141,
        "check_percent": 93.617,
        "wall_s": 189,
        "median_decode_tps": 177.7,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 6,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 4280,
            "min": 164.6,
            "median": 177.7,
            "mean": 178.33333333333334,
            "p95": 189.2,
            "max": 193
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 189,
            "min": 4.1,
            "median": 7.949999999999999,
            "mean": 7.875,
            "p95": 10.7,
            "max": 13.1
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29021,
            "min": 617,
            "median": 1194,
            "mean": 1209.2083333333333,
            "p95": 1778,
            "max": 2108
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 68,
          "checks_possible": 69,
          "check_percent": 98.551,
          "wall_s": 85.1,
          "median_decode_tps": 182.15,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 2189.6,
              "min": 173,
              "median": 182.14999999999998,
              "mean": 182.46666666666667,
              "p95": 193,
              "max": 193
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 85.1,
              "min": 4.1,
              "median": 6.15,
              "mean": 7.091666666666667,
              "p95": 13.1,
              "max": 13.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13513,
              "min": 617,
              "median": 963,
              "mean": 1126.0833333333333,
              "p95": 2108,
              "max": 2108
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 64,
          "checks_possible": 72,
          "check_percent": 88.889,
          "wall_s": 103.89999999999999,
          "median_decode_tps": 174.9,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 2090.4,
              "min": 164.6,
              "median": 174.89999999999998,
              "mean": 174.2,
              "p95": 182.1,
              "max": 182.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 103.89999999999999,
              "min": 6.6,
              "median": 8.55,
              "mean": 8.658333333333333,
              "p95": 10.7,
              "max": 10.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15508,
              "min": 950,
              "median": 1237.5,
              "mean": 1292.3333333333333,
              "p95": 1631,
              "max": 1631
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-61",
      "run": "outputs/27b-q2-lab system-vulkan.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio 2.37.0 Vulkan",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 95.105,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 113.55,
      "ttft": null,
      "latency": 11.9,
      "wall": 301.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 5090 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core Ultra · RTX 5090",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 136,
        "checks_possible": 143,
        "check_percent": 95.105,
        "wall_s": 301.2,
        "median_decode_tps": 113.55,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 2711.4,
            "min": 102.4,
            "median": 113.55,
            "mean": 112.975,
            "p95": 120.7,
            "max": 122
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 301.2,
            "min": 6.3,
            "median": 11.9,
            "mean": 12.55,
            "p95": 20.2,
            "max": 22.2
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29167,
            "min": 617,
            "median": 1172,
            "mean": 1215.2916666666667,
            "p95": 1787,
            "max": 2159
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 69,
          "check_percent": 100,
          "wall_s": 139.1,
          "median_decode_tps": 115.55,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 1395.8,
              "min": 109.1,
              "median": 115.55000000000001,
              "mean": 116.31666666666666,
              "p95": 122,
              "max": 122
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 139.1,
              "min": 6.3,
              "median": 9.85,
              "mean": 11.591666666666667,
              "p95": 22.2,
              "max": 22.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13732,
              "min": 617,
              "median": 998.5,
              "mean": 1144.3333333333333,
              "p95": 2159,
              "max": 2159
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 74,
          "check_percent": 90.541,
          "wall_s": 162.1,
          "median_decode_tps": 109.45,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 1315.6,
              "min": 102.4,
              "median": 109.44999999999999,
              "mean": 109.63333333333333,
              "p95": 115.6,
              "max": 115.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 162.1,
              "min": 10.6,
              "median": 12.75,
              "mean": 13.508333333333333,
              "p95": 17.7,
              "max": 17.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15435,
              "min": 977,
              "median": 1204,
              "mean": 1286.25,
              "p95": 1690,
              "max": 1690
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-62",
      "run": "outputs/27b-q2-lab system.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio 2.37.0 CUDA 12, Windows",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.617,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 177.85,
      "ttft": null,
      "latency": 7.949999999999999,
      "wall": 188.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 5090 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core Ultra · RTX 5090",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 132,
        "checks_possible": 141,
        "check_percent": 93.617,
        "wall_s": 188.9,
        "median_decode_tps": 177.85,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 6,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 4280.2,
            "min": 165.2,
            "median": 177.85,
            "mean": 178.34166666666667,
            "p95": 189.2,
            "max": 192.9
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 188.9,
            "min": 4.1,
            "median": 7.949999999999999,
            "mean": 7.870833333333334,
            "p95": 10.6,
            "max": 13
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 28987,
            "min": 617,
            "median": 1194,
            "mean": 1207.7916666666667,
            "p95": 1744,
            "max": 2108
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 68,
          "checks_possible": 69,
          "check_percent": 98.551,
          "wall_s": 85.2,
          "median_decode_tps": 182.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 2189.4,
              "min": 173.2,
              "median": 182.1,
              "mean": 182.45,
              "p95": 192.9,
              "max": 192.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 85.2,
              "min": 4.1,
              "median": 6.1,
              "mean": 7.1,
              "p95": 13,
              "max": 13
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13479,
              "min": 617,
              "median": 963,
              "mean": 1123.25,
              "p95": 2108,
              "max": 2108
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 64,
          "checks_possible": 72,
          "check_percent": 88.889,
          "wall_s": 103.7,
          "median_decode_tps": 174.7,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 2090.8,
              "min": 165.2,
              "median": 174.7,
              "mean": 174.23333333333335,
              "p95": 181.9,
              "max": 181.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 103.7,
              "min": 6.6,
              "median": 8.55,
              "mean": 8.641666666666666,
              "p95": 10.6,
              "max": 10.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15508,
              "min": 950,
              "median": 1237.5,
              "mean": 1292.3333333333333,
              "p95": 1631,
              "max": 1631
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-63",
      "run": "outputs/27b-q2-lab system-llmster-nocoopmat.WEDGED-4-1item-20260913T2032.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "wedged/partial",
      "cases": 1,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 1,
      "strict": null,
      "prefill": null,
      "decode": 18.7,
      "ttft": null,
      "latency": 128.4,
      "wall": 128.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": null,
        "checks_earned": 7,
        "checks_possible": 7,
        "check_percent": 100,
        "wall_s": 128.4,
        "median_decode_tps": 18.7,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 18.7,
            "min": 18.7,
            "median": 18.7,
            "mean": 18.7,
            "p95": 18.7,
            "max": 18.7
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 1,
            "total": 128.4,
            "min": 128.4,
            "median": 128.4,
            "mean": 128.4,
            "p95": 128.4,
            "max": 128.4
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 1,
            "total": 1787,
            "min": 1787,
            "median": 1787,
            "mean": 1787,
            "p95": 1787,
            "max": 1787
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": null,
          "checks_earned": 7,
          "checks_possible": 7,
          "check_percent": 100,
          "wall_s": 128.4,
          "median_decode_tps": 18.7,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 18.7,
              "min": 18.7,
              "median": 18.7,
              "mean": 18.7,
              "p95": 18.7,
              "max": 18.7
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 1,
              "total": 128.4,
              "min": 128.4,
              "median": 128.4,
              "mean": 128.4,
              "p95": 128.4,
              "max": 128.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 1,
              "total": 1787,
              "min": 1787,
              "median": 1787,
              "mean": 1787,
              "p95": 1787,
              "max": 1787
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-64",
      "run": "outputs/27b-q2-lab system-lms.WEDGED-2-4items-20260913T1644.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "wedged/partial",
      "cases": 4,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 4,
      "strict": null,
      "prefill": null,
      "decode": 18.5,
      "ttft": null,
      "latency": 71.94999999999999,
      "wall": 308.09999999999997,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 4,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 4,
        "ok": 4,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 24,
        "checks_possible": 24,
        "check_percent": 100,
        "wall_s": 308.09999999999997,
        "median_decode_tps": 18.5,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 4,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 4,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 4,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 4,
            "total": 73.3,
            "min": 17,
            "median": 18.5,
            "mean": 18.325,
            "p95": 19.3,
            "max": 19.3
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 4,
            "total": 308.09999999999997,
            "min": 61.1,
            "median": 71.94999999999999,
            "mean": 77.02499999999999,
            "p95": 103.1,
            "max": 103.1
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 4,
            "total": 5278,
            "min": 1109,
            "median": 1191,
            "mean": 1319.5,
            "p95": 1787,
            "max": 1787
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 4,
          "ok": 4,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 24,
          "checks_possible": 24,
          "check_percent": 100,
          "wall_s": 308.09999999999997,
          "median_decode_tps": 18.5,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 4,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 4,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 4,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 4,
              "total": 73.3,
              "min": 17,
              "median": 18.5,
              "mean": 18.325,
              "p95": 19.3,
              "max": 19.3
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 4,
              "total": 308.09999999999997,
              "min": 61.1,
              "median": 71.94999999999999,
              "mean": 77.02499999999999,
              "p95": 103.1,
              "max": 103.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 4,
              "total": 5278,
              "min": 1109,
              "median": 1191,
              "mean": 1319.5,
              "p95": 1787,
              "max": 1787
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-65",
      "run": "outputs/27b-q2-lab system-lms.WEDGED-3-2items-20260913T1909.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "wedged/partial",
      "cases": 2,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 2,
      "strict": null,
      "prefill": null,
      "decode": 18.65,
      "ttft": null,
      "latency": 83.5,
      "wall": 167,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 13,
        "checks_possible": 13,
        "check_percent": 100,
        "wall_s": 167,
        "median_decode_tps": 18.65,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 37.3,
            "min": 18.6,
            "median": 18.65,
            "mean": 18.65,
            "p95": 18.7,
            "max": 18.7
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 2,
            "total": 167,
            "min": 64.9,
            "median": 83.5,
            "mean": 83.5,
            "p95": 102.1,
            "max": 102.1
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 2,
            "total": 2924,
            "min": 1137,
            "median": 1462,
            "mean": 1462,
            "p95": 1787,
            "max": 1787
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 2,
          "ok": 2,
          "structure": 2,
          "strict_passes": null,
          "checks_earned": 13,
          "checks_possible": 13,
          "check_percent": 100,
          "wall_s": 167,
          "median_decode_tps": 18.65,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 2,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 2,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 2,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 2,
              "total": 37.3,
              "min": 18.6,
              "median": 18.65,
              "mean": 18.65,
              "p95": 18.7,
              "max": 18.7
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 2,
              "total": 167,
              "min": 64.9,
              "median": 83.5,
              "mean": 83.5,
              "p95": 102.1,
              "max": 102.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 2,
              "total": 2924,
              "min": 1137,
              "median": 1462,
              "mean": 1462,
              "p95": 1787,
              "max": 1787
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-66",
      "run": "outputs/27b-q2-lab system-lms.WEDGED-RUN-DISCARDED.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "frozen24",
      "status": "wedged/partial",
      "cases": 16,
      "expected": 24,
      "success": 62.5,
      "quality": 100,
      "structure": 10,
      "strict": null,
      "prefill": null,
      "decode": 18.6,
      "ttft": null,
      "latency": 91.75,
      "wall": 2512.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 16,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 16,
        "ok": 10,
        "structure": 10,
        "strict_passes": null,
        "checks_earned": 58,
        "checks_possible": 58,
        "check_percent": 100,
        "wall_s": 2512.4,
        "median_decode_tps": 18.6,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 6,
        "scored_cases": 16,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 10,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 16,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 10,
            "total": 185.1,
            "min": 17.1,
            "median": 18.6,
            "mean": 18.51,
            "p95": 19.4,
            "max": 19.4
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 16,
            "total": 2512.4,
            "min": 49.7,
            "median": 91.75,
            "mean": 157.025,
            "p95": 300.7,
            "max": 300.7
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 10,
            "total": 12107,
            "min": 850,
            "median": 1056,
            "mean": 1210.7,
            "p95": 2159,
            "max": 2159
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 10,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 58,
          "checks_possible": 58,
          "check_percent": 100,
          "wall_s": 1309.6,
          "median_decode_tps": 18.6,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 2,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 10,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 10,
              "total": 185.1,
              "min": 17.1,
              "median": 18.6,
              "mean": 18.51,
              "p95": 19.4,
              "max": 19.4
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1309.6,
              "min": 49.7,
              "median": 63.05,
              "mean": 109.13333333333333,
              "p95": 300.7,
              "max": 300.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 10,
              "total": 12107,
              "min": 850,
              "median": 1056,
              "mean": 1210.7,
              "p95": 2159,
              "max": 2159
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 4,
          "ok": 0,
          "structure": 0,
          "strict_passes": null,
          "checks_earned": 0,
          "checks_possible": 0,
          "check_percent": null,
          "wall_s": 1202.8,
          "median_decode_tps": null,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": null,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 4,
          "scored_cases": 4,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 0,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 4,
          "http_requests": null,
          "metrics": {
            "decode_tps": null,
            "prefill_tps": null,
            "elapsed_s": {
              "n": 4,
              "total": 1202.8,
              "min": 300.7,
              "median": 300.7,
              "mean": 300.7,
              "p95": 300.7,
              "max": 300.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": null,
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-67",
      "run": "outputs/27b-q4-b10920-pinned.jsonl",
      "model": "27B Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "SYCL b10920 (`chat-lab`), runtime PM pinned on, display C-states and ASPM off",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 76.056,
      "structure": 24,
      "strict": null,
      "prefill": null,
      "decode": 36.150000000000006,
      "ttft": null,
      "latency": 41.35,
      "wall": 1005.5,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 24,
        "strict_passes": null,
        "checks_earned": 108,
        "checks_possible": 142,
        "check_percent": 76.056,
        "wall_s": 1005.5,
        "median_decode_tps": 36.15,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 27,
        "fabricated_quotes": 7,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 865.4,
            "min": 34.2,
            "median": 36.150000000000006,
            "mean": 36.05833333333333,
            "p95": 36.9,
            "max": 37.2
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1005.5,
            "min": 25.8,
            "median": 41.35,
            "mean": 41.895833333333336,
            "p95": 65.2,
            "max": 66.1
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 33004,
            "min": 850,
            "median": 1323,
            "mean": 1375.1666666666667,
            "p95": 2135,
            "max": 2150
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 60,
          "checks_possible": 65,
          "check_percent": 92.308,
          "wall_s": 435.8,
          "median_decode_tps": 36.65,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 436.5,
              "min": 35.2,
              "median": 36.650000000000006,
              "mean": 36.375,
              "p95": 37.2,
              "max": 37.2
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 435.8,
              "min": 25.8,
              "median": 31.65,
              "mean": 36.31666666666666,
              "p95": 64.3,
              "max": 64.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14476,
              "min": 850,
              "median": 1049,
              "mean": 1206.3333333333333,
              "p95": 2150,
              "max": 2150
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 48,
          "checks_possible": 77,
          "check_percent": 62.338,
          "wall_s": 569.7,
          "median_decode_tps": 35.9,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 22,
          "fabricated_quotes": 7,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 428.9,
              "min": 34.2,
              "median": 35.9,
              "mean": 35.74166666666667,
              "p95": 36.9,
              "max": 36.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 569.7,
              "min": 37.7,
              "median": 44.8,
              "mean": 47.475,
              "p95": 66.1,
              "max": 66.1
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 18528,
              "min": 1205,
              "median": 1492,
              "mean": 1544,
              "p95": 2135,
              "max": 2135
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-68",
      "run": "outputs/27b-q4-b10920.jsonl",
      "model": "27B Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "SYCL b10920",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 78.014,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 35.25,
      "ttft": null,
      "latency": 41.3,
      "wall": 1030.5,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 110,
        "checks_possible": 141,
        "check_percent": 78.014,
        "wall_s": 1030.5,
        "median_decode_tps": 35.25,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 24,
        "fabricated_quotes": 7,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 840.4,
            "min": 33.6,
            "median": 35.25,
            "mean": 35.016666666666666,
            "p95": 36.1,
            "max": 36.1
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1030.5,
            "min": 28,
            "median": 41.3,
            "mean": 42.9375,
            "p95": 66.5,
            "max": 67.2
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 32844,
            "min": 885,
            "median": 1339.5,
            "mean": 1368.5,
            "p95": 2127,
            "max": 2150
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 60,
          "checks_possible": 65,
          "check_percent": 92.308,
          "wall_s": 451.6,
          "median_decode_tps": 35.4,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 421.7,
              "min": 34,
              "median": 35.4,
              "mean": 35.141666666666666,
              "p95": 36.1,
              "max": 36.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 451.6,
              "min": 28,
              "median": 33.35,
              "mean": 37.63333333333333,
              "p95": 66.5,
              "max": 66.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14483,
              "min": 885,
              "median": 1038,
              "mean": 1206.9166666666667,
              "p95": 2150,
              "max": 2150
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 50,
          "checks_possible": 76,
          "check_percent": 65.789,
          "wall_s": 578.9,
          "median_decode_tps": 34.95,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 19,
          "fabricated_quotes": 7,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 418.7,
              "min": 33.6,
              "median": 34.95,
              "mean": 34.891666666666666,
              "p95": 36.1,
              "max": 36.1
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 578.9,
              "min": 39.6,
              "median": 46,
              "mean": 48.24166666666667,
              "p95": 67.2,
              "max": 67.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 18361,
              "min": 1285,
              "median": 1471,
              "mean": 1530.0833333333333,
              "p95": 2127,
              "max": 2127
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-69",
      "run": "outputs/27b-q4-b10964-pinned.jsonl",
      "model": "27B Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "SYCL b10964 with `chat-lab`'s arguments, same pins",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 76.596,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 36.25,
      "ttft": null,
      "latency": 39.6,
      "wall": 995.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 108,
        "checks_possible": 141,
        "check_percent": 76.596,
        "wall_s": 995.4,
        "median_decode_tps": 36.25,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 26,
        "fabricated_quotes": 7,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 866.7,
            "min": 34.8,
            "median": 36.25,
            "mean": 36.1125,
            "p95": 36.9,
            "max": 37.2
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 995.4,
            "min": 27.3,
            "median": 39.6,
            "mean": 41.475,
            "p95": 64.3,
            "max": 65.3
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 32651,
            "min": 915,
            "median": 1296.5,
            "mean": 1360.4583333333333,
            "p95": 2135,
            "max": 2150
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 60,
          "checks_possible": 65,
          "check_percent": 92.308,
          "wall_s": 439.6,
          "median_decode_tps": 36.7,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 437.5,
              "min": 35.2,
              "median": 36.7,
              "mean": 36.458333333333336,
              "p95": 37.2,
              "max": 37.2
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 439.6,
              "min": 27.3,
              "median": 31.5,
              "mean": 36.63333333333333,
              "p95": 64.3,
              "max": 64.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14601,
              "min": 915,
              "median": 1044,
              "mean": 1216.75,
              "p95": 2150,
              "max": 2150
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 48,
          "checks_possible": 76,
          "check_percent": 63.158,
          "wall_s": 555.8,
          "median_decode_tps": 35.9,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 21,
          "fabricated_quotes": 7,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 429.2,
              "min": 34.8,
              "median": 35.9,
              "mean": 35.766666666666666,
              "p95": 36.6,
              "max": 36.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 555.8,
              "min": 36.6,
              "median": 43.3,
              "mean": 46.31666666666667,
              "p95": 65.3,
              "max": 65.3
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 18050,
              "min": 1212,
              "median": 1412,
              "mean": 1504.1666666666667,
              "p95": 2135,
              "max": 2135
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-70",
      "run": "outputs/27b-q4-b10964-rp1.1-pinned.jsonl",
      "model": "27B Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "as `27b-q4-b10964-pinned`, plus `--repeat-penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.151,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 34.7,
      "ttft": null,
      "latency": 42.349999999999994,
      "wall": 1046.7,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 136,
        "checks_possible": 146,
        "check_percent": 93.151,
        "wall_s": 1046.7,
        "median_decode_tps": 34.7,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 6,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 835.8,
            "min": 33,
            "median": 34.7,
            "mean": 34.825,
            "p95": 36.4,
            "max": 36.6
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1046.7,
            "min": 25.6,
            "median": 42.349999999999994,
            "mean": 43.6125,
            "p95": 64.8,
            "max": 68.9
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 33248,
            "min": 822,
            "median": 1364.5,
            "mean": 1385.3333333333333,
            "p95": 2041,
            "max": 2188
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 63,
          "checks_possible": 67,
          "check_percent": 94.03,
          "wall_s": 456.4,
          "median_decode_tps": 34.95,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 3,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 421.5,
              "min": 33.9,
              "median": 34.95,
              "mean": 35.125,
              "p95": 36.6,
              "max": 36.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 456.4,
              "min": 25.6,
              "median": 34.95,
              "mean": 38.03333333333333,
              "p95": 64,
              "max": 64
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14626,
              "min": 822,
              "median": 1128.5,
              "mean": 1218.8333333333333,
              "p95": 1980,
              "max": 1980
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 73,
          "checks_possible": 79,
          "check_percent": 92.405,
          "wall_s": 590.3,
          "median_decode_tps": 34.55,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 414.3,
              "min": 33,
              "median": 34.55,
              "mean": 34.525,
              "p95": 35.7,
              "max": 35.7
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 590.3,
              "min": 38.4,
              "median": 47.5,
              "mean": 49.19166666666666,
              "p95": 68.9,
              "max": 68.9
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 18622,
              "min": 1173,
              "median": 1466.5,
              "mean": 1551.8333333333333,
              "p95": 2188,
              "max": 2188
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-71",
      "run": "outputs/9b-fast-lab.jsonl",
      "model": "Qwen3.5-9B Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "`fast-a770`: SYCL b10920, 32k, 2 slots, q8_0 KV, as served (`repeat_penalty 1.0`)",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 71.756,
      "structure": 20,
      "strict": null,
      "prefill": null,
      "decode": 25.55,
      "ttft": null,
      "latency": 55.8,
      "wall": 1565.4,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": 32768,
      "slots": 2,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc A770",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 20,
        "strict_passes": null,
        "checks_earned": 94,
        "checks_possible": 131,
        "check_percent": 71.756,
        "wall_s": 1565.4,
        "median_decode_tps": 25.55,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 30,
        "fabricated_quotes": 7,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 606.4,
            "min": 24.2,
            "median": 25.55,
            "mean": 25.266666666666666,
            "p95": 26.1,
            "max": 26.2
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1565.4,
            "min": 33.8,
            "median": 55.8,
            "mean": 65.225,
            "p95": 122.9,
            "max": 165.7
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 37968,
            "min": 855,
            "median": 1346,
            "mean": 1582,
            "p95": 2921,
            "max": 3943
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 50,
          "checks_possible": 61,
          "check_percent": 81.967,
          "wall_s": 647.5,
          "median_decode_tps": 25.75,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 11,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 306.7,
              "min": 24.2,
              "median": 25.75,
              "mean": 25.558333333333334,
              "p95": 26.2,
              "max": 26.2
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 647.5,
              "min": 33.8,
              "median": 49.150000000000006,
              "mean": 53.958333333333336,
              "p95": 91.2,
              "max": 91.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15876,
              "min": 855,
              "median": 1219.5,
              "mean": 1323,
              "p95": 2241,
              "max": 2241
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 9,
          "strict_passes": null,
          "checks_earned": 44,
          "checks_possible": 70,
          "check_percent": 62.857,
          "wall_s": 917.9,
          "median_decode_tps": 25.15,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 19,
          "fabricated_quotes": 7,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 299.7,
              "min": 24.2,
              "median": 25.15,
              "mean": 24.975,
              "p95": 25.8,
              "max": 25.8
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 917.9,
              "min": 52.7,
              "median": 63,
              "mean": 76.49166666666666,
              "p95": 165.7,
              "max": 165.7
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 22092,
              "min": 1232,
              "median": 1524,
              "mean": 1841,
              "p95": 3943,
              "max": 3943
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-72",
      "run": "outputs/flash-next-q2.jsonl",
      "model": "Flash-Next MoE Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "CUDA b10820",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 89.231,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 15.75,
      "ttft": null,
      "latency": 175.6,
      "wall": 5040.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 2×4060 Ti + Xeon",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 116,
        "checks_possible": 130,
        "check_percent": 89.231,
        "wall_s": 5040.3,
        "median_decode_tps": 15.75,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 8,
        "fabricated_quotes": 6,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 366.2,
            "min": 5,
            "median": 15.75,
            "mean": 15.258333333333333,
            "p95": 17.5,
            "max": 18.9
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 5040.3,
            "min": 83.1,
            "median": 175.6,
            "mean": 210.01250000000002,
            "p95": 380.6,
            "max": 733.6
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 33764,
            "min": 744,
            "median": 1218.5,
            "mean": 1406.8333333333333,
            "p95": 2085,
            "max": 3931
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 63,
          "check_percent": 98.413,
          "wall_s": 2467.9,
          "median_decode_tps": 16.15,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 183.4,
              "min": 5,
              "median": 16.15,
              "mean": 15.283333333333333,
              "p95": 17.4,
              "max": 17.4
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 2467.9,
              "min": 83.1,
              "median": 148.55,
              "mean": 205.65833333333333,
              "p95": 733.6,
              "max": 733.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13801,
              "min": 744,
              "median": 965,
              "mean": 1150.0833333333333,
              "p95": 1930,
              "max": 1930
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 54,
          "checks_possible": 67,
          "check_percent": 80.597,
          "wall_s": 2572.4,
          "median_decode_tps": 14.9,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 8,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 182.8,
              "min": 13.2,
              "median": 14.899999999999999,
              "mean": 15.233333333333333,
              "p95": 18.9,
              "max": 18.9
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 2572.4,
              "min": 86.4,
              "median": 224.15,
              "mean": 214.36666666666667,
              "p95": 380.6,
              "max": 380.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 19963,
              "min": 1026,
              "median": 1516,
              "mean": 1663.5833333333333,
              "p95": 3931,
              "max": 3931
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-73",
      "run": "outputs/flash-next-q4.jsonl",
      "model": "Flash-Next MoE Q4_K_XL",
      "model_family": "Flash Next",
      "quant": "Q4_K_XL",
      "runtime": "CUDA b10820",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 87.805,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 10.649999999999999,
      "ttft": null,
      "latency": 216.65,
      "wall": 5790.2,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 2×4060 Ti + Xeon",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 108,
        "checks_possible": 123,
        "check_percent": 87.805,
        "wall_s": 5790.2,
        "median_decode_tps": 10.65,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 6,
        "fabricated_quotes": 9,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 255,
            "min": 9.3,
            "median": 10.649999999999999,
            "mean": 10.625,
            "p95": 11.5,
            "max": 12.6
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 5790.2,
            "min": 110.1,
            "median": 216.65,
            "mean": 241.25833333333333,
            "p95": 385.2,
            "max": 460.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 33745,
            "min": 731,
            "median": 1262,
            "mean": 1406.0416666666667,
            "p95": 2001,
            "max": 3896
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 58,
          "checks_possible": 60,
          "check_percent": 96.667,
          "wall_s": 2434.3,
          "median_decode_tps": 10.7,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 129.1,
              "min": 10.1,
              "median": 10.7,
              "mean": 10.758333333333333,
              "p95": 11.5,
              "max": 11.5
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 2434.3,
              "min": 110.1,
              "median": 180,
              "mean": 202.85833333333335,
              "p95": 343,
              "max": 343
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 13655,
              "min": 731,
              "median": 1016,
              "mean": 1137.9166666666667,
              "p95": 2001,
              "max": 2001
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 50,
          "checks_possible": 63,
          "check_percent": 79.365,
          "wall_s": 3355.9,
          "median_decode_tps": 10.35,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 8,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 125.89999999999999,
              "min": 9.3,
              "median": 10.35,
              "mean": 10.491666666666667,
              "p95": 12.6,
              "max": 12.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 3355.9,
              "min": 125.5,
              "median": 287.4,
              "mean": 279.65833333333336,
              "p95": 460.5,
              "max": 460.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 20090,
              "min": 1035,
              "median": 1452.5,
              "mean": 1674.1666666666667,
              "p95": 3896,
              "max": 3896
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-74",
      "run": "outputs/moe-b10920.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "frozen24",
      "status": "partial/probe",
      "cases": 3,
      "expected": 24,
      "success": 100,
      "quality": 100,
      "structure": 3,
      "strict": null,
      "prefill": null,
      "decode": 14.2,
      "ttft": null,
      "latency": 189.2,
      "wall": 664.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 3,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 3,
        "ok": 3,
        "structure": 3,
        "strict_passes": null,
        "checks_earned": 17,
        "checks_possible": 17,
        "check_percent": 100,
        "wall_s": 664.6,
        "median_decode_tps": 14.2,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 3,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 3,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 3,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 3,
            "total": 42.9,
            "min": 14.1,
            "median": 14.2,
            "mean": 14.299999999999999,
            "p95": 14.6,
            "max": 14.6
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 3,
            "total": 664.6,
            "min": 163.6,
            "median": 189.2,
            "mean": 221.53333333333333,
            "p95": 311.8,
            "max": 311.8
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 3,
            "total": 3961,
            "min": 1165,
            "median": 1205,
            "mean": 1320.3333333333333,
            "p95": 1591,
            "max": 1591
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": null,
          "checks_earned": 17,
          "checks_possible": 17,
          "check_percent": 100,
          "wall_s": 664.6,
          "median_decode_tps": 14.2,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 42.9,
              "min": 14.1,
              "median": 14.2,
              "mean": 14.299999999999999,
              "p95": 14.6,
              "max": 14.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 3,
              "total": 664.6,
              "min": 163.6,
              "median": 189.2,
              "mean": 221.53333333333333,
              "p95": 311.8,
              "max": 311.8
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 3,
              "total": 3961,
              "min": 1165,
              "median": 1205,
              "mean": 1320.3333333333333,
              "p95": 1591,
              "max": 1591
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-75",
      "run": "outputs/qwen3.6-35b-a3b.jsonl",
      "model": "35B-A3B IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "CUDA b10820",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 59.434,
      "structure": 16,
      "strict": null,
      "prefill": null,
      "decode": 74.85,
      "ttft": null,
      "latency": 20.25,
      "wall": 482.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 2×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 16,
        "strict_passes": null,
        "checks_earned": 63,
        "checks_possible": 106,
        "check_percent": 59.434,
        "wall_s": 482.9,
        "median_decode_tps": 74.85,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 28,
        "fabricated_quotes": 15,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1789.6,
            "min": 73.1,
            "median": 74.85,
            "mean": 74.56666666666666,
            "p95": 75.4,
            "max": 75.6
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 482.9,
            "min": 8.6,
            "median": 20.25,
            "mean": 20.120833333333334,
            "p95": 29.8,
            "max": 32.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 34155,
            "min": 596,
            "median": 1439.5,
            "mean": 1423.125,
            "p95": 2102,
            "max": 2278
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 48,
          "checks_possible": 65,
          "check_percent": 73.846,
          "wall_s": 202.3,
          "median_decode_tps": 75.05,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 12,
          "fabricated_quotes": 5,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 898.4,
              "min": 73.4,
              "median": 75.05,
              "mean": 74.86666666666666,
              "p95": 75.6,
              "max": 75.6
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 202.3,
              "min": 8.6,
              "median": 14.850000000000001,
              "mean": 16.858333333333334,
              "p95": 29.6,
              "max": 29.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14343,
              "min": 596,
              "median": 1053.5,
              "mean": 1195.25,
              "p95": 2062,
              "max": 2062
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 15,
          "checks_possible": 41,
          "check_percent": 36.585,
          "wall_s": 280.6,
          "median_decode_tps": 74.35,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 16,
          "fabricated_quotes": 10,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 891.2,
              "min": 73.1,
              "median": 74.35,
              "mean": 74.26666666666667,
              "p95": 75,
              "max": 75
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 280.6,
              "min": 16.7,
              "median": 23.15,
              "mean": 23.383333333333333,
              "p95": 32.5,
              "max": 32.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 19812,
              "min": 1190,
              "median": 1639.5,
              "mean": 1651,
              "p95": 2278,
              "max": 2278
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-76",
      "run": "outputs/qwen3.8-27b-q2.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "SYCL b10644, production flags, MTP",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 81.208,
      "structure": 22,
      "strict": null,
      "prefill": null,
      "decode": 14.1,
      "ttft": null,
      "latency": 92.4,
      "wall": 2259.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 121,
        "checks_possible": 149,
        "check_percent": 81.208,
        "wall_s": 2259.6,
        "median_decode_tps": 14.1,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 24,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 337.1,
            "min": 13.1,
            "median": 14.1,
            "mean": 14.045833333333333,
            "p95": 14.4,
            "max": 14.5
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 2259.6,
            "min": 49.2,
            "median": 92.4,
            "mean": 94.15,
            "p95": 134.7,
            "max": 155.8
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 29609,
            "min": 641,
            "median": 1237.5,
            "mean": 1233.7083333333333,
            "p95": 1834,
            "max": 2085
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 71,
          "check_percent": 97.183,
          "wall_s": 1057.6,
          "median_decode_tps": 14.25,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 2,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 170.2,
              "min": 13.6,
              "median": 14.25,
              "mean": 14.183333333333334,
              "p95": 14.4,
              "max": 14.4
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1057.6,
              "min": 49.2,
              "median": 74,
              "mean": 88.13333333333333,
              "p95": 155.8,
              "max": 155.8
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14089,
              "min": 641,
              "median": 991,
              "mean": 1174.0833333333333,
              "p95": 2085,
              "max": 2085
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 52,
          "checks_possible": 78,
          "check_percent": 66.667,
          "wall_s": 1202,
          "median_decode_tps": 14,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 22,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 166.9,
              "min": 13.1,
              "median": 14,
              "mean": 13.908333333333333,
              "p95": 14.5,
              "max": 14.5
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1202,
              "min": 84.3,
              "median": 99.25,
              "mean": 100.16666666666667,
              "p95": 123.4,
              "max": 123.4
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 15520,
              "min": 1096,
              "median": 1282.5,
              "mean": 1293.3333333333333,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-77",
      "run": "outputs/qwen3.8-27b-q8.jsonl",
      "model": "27B Q8_0",
      "model_family": "27B",
      "quant": "Q8_0",
      "runtime": "CUDA b10820, no speculation",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 76.224,
      "structure": 23,
      "strict": null,
      "prefill": null,
      "decode": 9.4,
      "ttft": null,
      "latency": 134.95,
      "wall": 3481.6,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 2×4060 Ti",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 23,
        "strict_passes": null,
        "checks_earned": 109,
        "checks_possible": 143,
        "check_percent": 76.224,
        "wall_s": 3481.6,
        "median_decode_tps": 9.4,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 29,
        "fabricated_quotes": 5,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 226,
            "min": 9.4,
            "median": 9.4,
            "mean": 9.416666666666668,
            "p95": 9.5,
            "max": 9.5
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 3481.6,
            "min": 90.7,
            "median": 134.95,
            "mean": 145.06666666666666,
            "p95": 206.2,
            "max": 218
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 32249,
            "min": 843,
            "median": 1243,
            "mean": 1343.7083333333333,
            "p95": 1907,
            "max": 2013
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 62,
          "checks_possible": 67,
          "check_percent": 92.537,
          "wall_s": 1570.5,
          "median_decode_tps": 9.4,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 1,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 113.2,
              "min": 9.4,
              "median": 9.4,
              "mean": 9.433333333333334,
              "p95": 9.5,
              "max": 9.5
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1570.5,
              "min": 90.7,
              "median": 116.8,
              "mean": 130.875,
              "p95": 218,
              "max": 218
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14562,
              "min": 843,
              "median": 1085,
              "mean": 1213.5,
              "p95": 2013,
              "max": 2013
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 11,
          "strict_passes": null,
          "checks_earned": 47,
          "checks_possible": 76,
          "check_percent": 61.842,
          "wall_s": 1911.1,
          "median_decode_tps": 9.4,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 25,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 112.80000000000001,
              "min": 9.4,
              "median": 9.4,
              "mean": 9.4,
              "p95": 9.4,
              "max": 9.4
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 1911.1,
              "min": 128.5,
              "median": 158.5,
              "mean": 159.25833333333333,
              "p95": 206.2,
              "max": 206.2
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 17687,
              "min": 1188,
              "median": 1476.5,
              "mean": 1473.9166666666667,
              "p95": 1907,
              "max": 1907
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-78",
      "run": "outputs/qwen3.8-27b.jsonl",
      "model": "27B Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "SYCL b10644, production flags",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 76.761,
      "structure": 24,
      "strict": null,
      "prefill": null,
      "decode": 34.55,
      "ttft": null,
      "latency": 41.849999999999994,
      "wall": 1035.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 24,
        "strict_passes": null,
        "checks_earned": 109,
        "checks_possible": 142,
        "check_percent": 76.761,
        "wall_s": 1035.3,
        "median_decode_tps": 34.55,
        "median_prefill_tps": null,
        "retries": null,
        "truncated": 0,
        "near_quotes": 25,
        "fabricated_quotes": 8,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 828.7,
            "min": 33.2,
            "median": 34.55,
            "mean": 34.52916666666667,
            "p95": 35.3,
            "max": 35.4
          },
          "prefill_tps": null,
          "elapsed_s": {
            "n": 24,
            "total": 1035.3,
            "min": 27.8,
            "median": 41.849999999999994,
            "mean": 43.1375,
            "p95": 66.6,
            "max": 67.5
          },
          "ttft_s": null,
          "prompt_tokens": null,
          "completion_tokens": {
            "n": 24,
            "total": 32800,
            "min": 888,
            "median": 1304,
            "mean": 1366.6666666666667,
            "p95": 2132,
            "max": 2150
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 61,
          "checks_possible": 66,
          "check_percent": 92.424,
          "wall_s": 460.8,
          "median_decode_tps": 35.1,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 417.5,
              "min": 33.7,
              "median": 35.1,
              "mean": 34.791666666666664,
              "p95": 35.4,
              "max": 35.4
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 460.8,
              "min": 27.8,
              "median": 33.95,
              "mean": 38.4,
              "p95": 66.6,
              "max": 66.6
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 14781,
              "min": 888,
              "median": 1054,
              "mean": 1231.75,
              "p95": 2150,
              "max": 2150
            },
            "cached_prompt_tokens": null
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 48,
          "checks_possible": 76,
          "check_percent": 63.158,
          "wall_s": 574.5,
          "median_decode_tps": 34.4,
          "median_prefill_tps": null,
          "retries": null,
          "truncated": 0,
          "near_quotes": 20,
          "fabricated_quotes": 8,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 411.2,
              "min": 33.2,
              "median": 34.4,
              "mean": 34.266666666666666,
              "p95": 35.2,
              "max": 35.2
            },
            "prefill_tps": null,
            "elapsed_s": {
              "n": 12,
              "total": 574.5,
              "min": 36.5,
              "median": 44.9,
              "mean": 47.875,
              "p95": 67.5,
              "max": 67.5
            },
            "ttft_s": null,
            "prompt_tokens": null,
            "completion_tokens": {
              "n": 12,
              "total": 18019,
              "min": 1164,
              "median": 1399.5,
              "mean": 1501.5833333333333,
              "p95": 2132,
              "max": 2132
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-79",
      "run": "prod-acceptance/prod-acceptance-b10820.jsonl",
      "model": "27B Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "production `chat-27b-q2`: CUDA b10820, 49k ctx, 2 slots, q8_0 KV, MTP, `--repeat-penalty 1.1`",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 93.056,
      "structure": 22,
      "strict": null,
      "prefill": 685.3473781225991,
      "decode": 42.006831724971754,
      "ttft": null,
      "latency": 32.599999999999994,
      "wall": 785.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": 49152,
      "slots": 2,
      "kv": "q8_0",
      "date": "historical · date not in filename",
      "duplicate_of": "[private path omitted]",
      "unique_cases": 24,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived run inventory: lab system, 1×4060 Ti",
      "notes": "Exact duplicate artifact of [private path omitted]",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 134,
        "checks_possible": 144,
        "check_percent": 93.056,
        "wall_s": 785.8,
        "median_decode_tps": 42.007,
        "median_prefill_tps": 685.347,
        "retries": null,
        "truncated": 0,
        "near_quotes": 7,
        "fabricated_quotes": 3,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1003.577372235679,
            "min": 39.81517576244966,
            "median": 42.006831724971754,
            "mean": 41.81572384315329,
            "p95": 43.47408332237555,
            "max": 45.224965632758405
          },
          "prefill_tps": {
            "n": 24,
            "total": 16401.60723050997,
            "min": 635.5594246273733,
            "median": 685.3473781225991,
            "mean": 683.4003012712487,
            "p95": 721.6646397690673,
            "max": 772.9045917273872
          },
          "elapsed_s": {
            "n": 24,
            "total": 785.8,
            "min": 16.8,
            "median": 32.599999999999994,
            "mean": 32.74166666666667,
            "p95": 45.1,
            "max": 55.9
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 28820,
            "min": 617,
            "median": 1173,
            "mean": 1200.8333333333333,
            "p95": 1758,
            "max": 2152
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 68,
          "check_percent": 98.529,
          "wall_s": 352.7,
          "median_decode_tps": 42.66,
          "median_prefill_tps": 689.108,
          "retries": null,
          "truncated": 0,
          "near_quotes": 1,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 513.8887110992708,
              "min": 41.90179457359517,
              "median": 42.66018957495332,
              "mean": 42.82405925827257,
              "p95": 45.224965632758405,
              "max": 45.224965632758405
            },
            "prefill_tps": {
              "n": 12,
              "total": 8304.638243733618,
              "min": 635.5594246273733,
              "median": 689.1081634267066,
              "mean": 692.0531869778015,
              "p95": 772.9045917273872,
              "max": 772.9045917273872
            },
            "elapsed_s": {
              "n": 12,
              "total": 352.7,
              "min": 16.8,
              "median": 25.299999999999997,
              "mean": 29.391666666666666,
              "p95": 55.9,
              "max": 55.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13430,
              "min": 617,
              "median": 972.5,
              "mean": 1119.1666666666667,
              "p95": 2152,
              "max": 2152
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 67,
          "checks_possible": 76,
          "check_percent": 88.158,
          "wall_s": 433.1,
          "median_decode_tps": 40.575,
          "median_prefill_tps": 681.581,
          "retries": null,
          "truncated": 0,
          "near_quotes": 6,
          "fabricated_quotes": 3,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 489.68866113640814,
              "min": 39.81517576244966,
              "median": 40.5750570237494,
              "mean": 40.80738842803401,
              "p95": 42.67400380534174,
              "max": 42.67400380534174
            },
            "prefill_tps": {
              "n": 12,
              "total": 8096.968986776352,
              "min": 641.317203101215,
              "median": 681.5813872152182,
              "mean": 674.747415564696,
              "p95": 691.7811347672581,
              "max": 691.7811347672581
            },
            "elapsed_s": {
              "n": 12,
              "total": 433.1,
              "min": 30.8,
              "median": 34.5,
              "mean": 36.09166666666667,
              "p95": 45.1,
              "max": 45.1
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15390,
              "min": 1006,
              "median": 1250,
              "mean": 1282.5,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-80",
      "run": "prod-acceptance/prod-acceptance.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "frozen24",
      "status": "complete",
      "cases": 24,
      "expected": 24,
      "success": 100,
      "quality": 94.483,
      "structure": 22,
      "strict": null,
      "prefill": 689.1537756619912,
      "decode": 42.181510206755796,
      "ttft": null,
      "latency": 33.6,
      "wall": 803.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": 0,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": "[private path omitted]",
      "unique_cases": 24,
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Exact duplicate artifact of [private path omitted]",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 24,
        "ok": 24,
        "structure": 22,
        "strict_passes": null,
        "checks_earned": 137,
        "checks_possible": 145,
        "check_percent": 94.483,
        "wall_s": 803.1,
        "median_decode_tps": 42.182,
        "median_prefill_tps": 689.154,
        "retries": null,
        "truncated": 0,
        "near_quotes": 4,
        "fabricated_quotes": 4,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 24,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 24,
        "cached_prompt_tokens_recorded_cases": 24,
        "elapsed_s_recorded_cases": 24,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 24,
            "total": 1009.1913297781709,
            "min": 39.52686492222267,
            "median": 42.181510206755796,
            "mean": 42.04963874075712,
            "p95": 44.70712219125691,
            "max": 45.33654741166923
          },
          "prefill_tps": {
            "n": 24,
            "total": 16615.753208689766,
            "min": 660.6999517076573,
            "median": 689.1537756619912,
            "mean": 692.3230503620737,
            "p95": 738.1404859045164,
            "max": 777.8423705360327
          },
          "elapsed_s": {
            "n": 24,
            "total": 803.1,
            "min": 16.8,
            "median": 33.6,
            "mean": 33.4625,
            "p95": 46.7,
            "max": 55.3
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 24,
            "total": 61464,
            "min": 1219,
            "median": 2191,
            "mean": 2561,
            "p95": 4337,
            "max": 4936
          },
          "completion_tokens": {
            "n": 24,
            "total": 29720,
            "min": 617,
            "median": 1191,
            "mean": 1238.3333333333333,
            "p95": 1780,
            "max": 2159
          },
          "cached_prompt_tokens": {
            "n": 24,
            "total": 0,
            "min": 0,
            "median": 0,
            "mean": 0,
            "p95": 0,
            "max": 0
          }
        }
      },
      "groups": {
        "consolidate": {
          "cases": 12,
          "ok": 12,
          "structure": 12,
          "strict_passes": null,
          "checks_earned": 68,
          "checks_possible": 68,
          "check_percent": 100,
          "wall_s": 359.09999999999997,
          "median_decode_tps": 43.067,
          "median_prefill_tps": 698.71,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 518.6929963748111,
              "min": 41.91071493692437,
              "median": 43.06726320430343,
              "mean": 43.22441636456759,
              "p95": 45.33654741166923,
              "max": 45.33654741166923
            },
            "prefill_tps": {
              "n": 12,
              "total": 8494.209144310107,
              "min": 660.6999517076573,
              "median": 698.7097970208159,
              "mean": 707.8507620258423,
              "p95": 777.8423705360327,
              "max": 777.8423705360327
            },
            "elapsed_s": {
              "n": 12,
              "total": 359.09999999999997,
              "min": 16.8,
              "median": 25.35,
              "mean": 29.925,
              "p95": 55.3,
              "max": 55.3
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 26032,
              "min": 1219,
              "median": 1906.5,
              "mean": 2169.3333333333335,
              "p95": 4337,
              "max": 4337
            },
            "completion_tokens": {
              "n": 12,
              "total": 13878,
              "min": 617,
              "median": 957.5,
              "mean": 1156.5,
              "p95": 2159,
              "max": 2159
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        },
        "distill": {
          "cases": 12,
          "ok": 12,
          "structure": 10,
          "strict_passes": null,
          "checks_earned": 69,
          "checks_possible": 77,
          "check_percent": 89.61,
          "wall_s": 444,
          "median_decode_tps": 40.46,
          "median_prefill_tps": 673.849,
          "retries": null,
          "truncated": 0,
          "near_quotes": 4,
          "fabricated_quotes": 4,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 12,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 12,
          "cached_prompt_tokens_recorded_cases": 12,
          "elapsed_s_recorded_cases": 12,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 12,
              "total": 490.4983334033599,
              "min": 39.52686492222267,
              "median": 40.460469566393584,
              "mean": 40.874861116946654,
              "p95": 42.761791746427136,
              "max": 42.761791746427136
            },
            "prefill_tps": {
              "n": 12,
              "total": 8121.54406437966,
              "min": 664.1070284187902,
              "median": 673.8485854880337,
              "mean": 676.7953386983049,
              "p95": 693.2975608457484,
              "max": 693.2975608457484
            },
            "elapsed_s": {
              "n": 12,
              "total": 444,
              "min": 30.3,
              "median": 36.05,
              "mean": 37,
              "p95": 46.7,
              "max": 46.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 12,
              "total": 35432,
              "min": 1834,
              "median": 2865.5,
              "mean": 2952.6666666666665,
              "p95": 4936,
              "max": 4936
            },
            "completion_tokens": {
              "n": 12,
              "total": 15842,
              "min": 1039,
              "median": 1329.5,
              "mean": 1320.1666666666667,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": {
              "n": 12,
              "total": 0,
              "min": 0,
              "median": 0,
              "mean": 0,
              "p95": 0,
              "max": 0
            }
          }
        }
      }
    },
    {
      "id": "run-81",
      "run": "20260917T154912Z/akm-model-eval-q2-131k-q4.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 91.837,
      "structure": 118,
      "strict": 82,
      "prefill": 529.8935670920955,
      "decode": 40.23,
      "ttft": null,
      "latency": 6.036,
      "wall": 2145.201,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260917T154912Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 118,
        "strict_passes": 82,
        "checks_earned": 675,
        "checks_possible": 735,
        "check_percent": 91.837,
        "wall_s": 2145.201,
        "median_decode_tps": 40.23,
        "median_prefill_tps": 529.894,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 4776.46,
            "min": 22.73,
            "median": 40.23,
            "mean": 38.8330081300813,
            "p95": 44.07,
            "max": 46.14
          },
          "prefill_tps": {
            "n": 123,
            "total": 64372.928217016866,
            "min": 222.94923058281836,
            "median": 529.8935670920955,
            "mean": 523.3571399757469,
            "p95": 689.9981578278582,
            "max": 728.9343832637207
          },
          "elapsed_s": {
            "n": 123,
            "total": 2145.201,
            "min": 1.379,
            "median": 6.036,
            "mean": 17.440658536585367,
            "p95": 77.436,
            "max": 116.443
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 63807,
            "min": 16,
            "median": 148,
            "mean": 518.7560975609756,
            "p95": 2169,
            "max": 4105
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 51,
          "strict_passes": 40,
          "checks_earned": 197,
          "checks_possible": 213,
          "check_percent": 92.488,
          "wall_s": 200.129,
          "median_decode_tps": 38.42,
          "median_prefill_tps": 444.273,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 2033.66,
              "min": 30.56,
              "median": 38.42,
              "mean": 39.10884615384615,
              "p95": 44.99,
              "max": 46.14
            },
            "prefill_tps": {
              "n": 52,
              "total": 22495.350823113164,
              "min": 222.94923058281836,
              "median": 444.2726181011318,
              "mean": 432.6029004444839,
              "p95": 585.9428034076147,
              "max": 665.2676987844571
            },
            "elapsed_s": {
              "n": 52,
              "total": 200.129,
              "min": 1.487,
              "median": 3.182,
              "mean": 3.8486346153846154,
              "p95": 8.613,
              "max": 9.566
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 6137,
              "min": 30,
              "median": 80.5,
              "mean": 118.01923076923077,
              "p95": 342,
              "max": 378
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 2,
          "checks_earned": 26,
          "checks_possible": 30,
          "check_percent": 86.667,
          "wall_s": 222.061,
          "median_decode_tps": 26.72,
          "median_prefill_tps": 580.5,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 76.69,
              "min": 22.73,
              "median": 26.72,
              "mean": 25.563333333333333,
              "p95": 27.24,
              "max": 27.24
            },
            "prefill_tps": {
              "n": 3,
              "total": 1734.7741307666652,
              "min": 534.6785142692489,
              "median": 580.5003522580582,
              "mean": 578.2580435888883,
              "p95": 619.5952642393579,
              "max": 619.5952642393579
            },
            "elapsed_s": {
              "n": 3,
              "total": 222.061,
              "min": 48.135,
              "median": 68.258,
              "mean": 74.02033333333333,
              "p95": 105.668,
              "max": 105.668
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 570,
              "min": 147,
              "median": 167,
              "mean": 190,
              "p95": 256,
              "max": 256
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 37,
          "strict_passes": 18,
          "checks_earned": 211,
          "checks_possible": 236,
          "check_percent": 89.407,
          "wall_s": 1531.2169999999999,
          "median_decode_tps": 41.35,
          "median_prefill_tps": 673.082,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1572.89,
              "min": 31.55,
              "median": 41.35,
              "mean": 40.33051282051282,
              "p95": 43.04,
              "max": 43.82
            },
            "prefill_tps": {
              "n": 39,
              "total": 26142.828371321553,
              "min": 605.6470304717128,
              "median": 673.0819667440762,
              "mean": 670.3289325979885,
              "p95": 704.974698017479,
              "max": 728.9343832637207
            },
            "elapsed_s": {
              "n": 39,
              "total": 1531.2169999999999,
              "min": 6.473,
              "median": 31.822,
              "mean": 39.261974358974356,
              "p95": 102.74,
              "max": 116.443
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 51788,
              "min": 54,
              "median": 1122,
              "mean": 1327.8974358974358,
              "p95": 3389,
              "max": 4105
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 27,
          "strict_passes": 22,
          "checks_earned": 241,
          "checks_possible": 256,
          "check_percent": 94.141,
          "wall_s": 191.794,
          "median_decode_tps": 39.02,
          "median_prefill_tps": 479.281,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1093.22,
              "min": 30.53,
              "median": 39.02,
              "mean": 37.69724137931034,
              "p95": 43.43,
              "max": 44.07
            },
            "prefill_tps": {
              "n": 29,
              "total": 13999.97489181549,
              "min": 280.5142761729838,
              "median": 479.2805406715313,
              "mean": 482.75775489018923,
              "p95": 680.4488488042658,
              "max": 683.8604342705207
            },
            "elapsed_s": {
              "n": 29,
              "total": 191.794,
              "min": 1.379,
              "median": 4.31,
              "mean": 6.613586206896552,
              "p95": 19.827,
              "max": 24.281
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 5312,
              "min": 16,
              "median": 129,
              "mean": 183.17241379310346,
              "p95": 699,
              "max": 952
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-82",
      "run": "20260917T165935Z/akm-model-eval-9b-131k-q8.jsonl",
      "model": "Qwen3.5-9B · Q4_K_M",
      "model_family": "9B",
      "quant": "Q4_K_M",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 84.626,
      "structure": 106,
      "strict": 60,
      "prefill": 1885.1733172659965,
      "decode": 40.77,
      "ttft": null,
      "latency": 4.714,
      "wall": 2409.482,
      "requests": 125,
      "retries": 0,
      "truncated": 2,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260917T165935Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 106,
        "strict_passes": 60,
        "checks_earned": 622,
        "checks_possible": 735,
        "check_percent": 84.626,
        "wall_s": 2409.482,
        "median_decode_tps": 40.77,
        "median_prefill_tps": 1885.173,
        "retries": 0,
        "truncated": 2,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 5007.81,
            "min": 35.41,
            "median": 40.77,
            "mean": 40.71390243902439,
            "p95": 41.53,
            "max": 41.81
          },
          "prefill_tps": {
            "n": 123,
            "total": 236796.12674592008,
            "min": 779.4496407753829,
            "median": 1885.1733172659965,
            "mean": 1925.1717621619518,
            "p95": 2637.58564211616,
            "max": 2674.246670284328
          },
          "elapsed_s": {
            "n": 123,
            "total": 2409.482,
            "min": 0.873,
            "median": 4.714,
            "mean": 19.589284552845527,
            "p95": 83.398,
            "max": 179.963
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 90784,
            "min": 8,
            "median": 172,
            "mean": 738.0813008130082,
            "p95": 3048,
            "max": 6000
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 33,
          "checks_earned": 185,
          "checks_possible": 213,
          "check_percent": 86.854,
          "wall_s": 200.868,
          "median_decode_tps": 40.615,
          "median_prefill_tps": 1562.503,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 2115.84,
              "min": 40.22,
              "median": 40.614999999999995,
              "mean": 40.68923076923077,
              "p95": 41.42,
              "max": 41.59
            },
            "prefill_tps": {
              "n": 52,
              "total": 78513.63783699443,
              "min": 779.4496407753829,
              "median": 1562.5032141134957,
              "mean": 1509.8776507114312,
              "p95": 2047.1442337938574,
              "max": 2304.852795534348
            },
            "elapsed_s": {
              "n": 52,
              "total": 200.868,
              "min": 1.134,
              "median": 2.532,
              "mean": 3.862846153846154,
              "p95": 10.208,
              "max": 14.32
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7674,
              "min": 35,
              "median": 91.5,
              "mean": 147.57692307692307,
              "p95": 414,
              "max": 582
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 1,
          "strict_passes": 0,
          "checks_earned": 18,
          "checks_possible": 30,
          "check_percent": 60,
          "wall_s": 422.639,
          "median_decode_tps": 36.75,
          "median_prefill_tps": 2160.336,
          "retries": 0,
          "truncated": 2,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 110.03,
              "min": 35.41,
              "median": 36.75,
              "mean": 36.67666666666666,
              "p95": 37.87,
              "max": 37.87
            },
            "prefill_tps": {
              "n": 3,
              "total": 6450.34684578443,
              "min": 1958.3486370152195,
              "median": 2160.3360049931293,
              "mean": 2150.1156152614767,
              "p95": 2331.662203776081,
              "max": 2331.662203776081
            },
            "elapsed_s": {
              "n": 3,
              "total": 422.639,
              "min": 72.801,
              "median": 169.875,
              "mean": 140.87966666666665,
              "p95": 179.963,
              "max": 179.963
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 13654,
              "min": 1654,
              "median": 6000,
              "mean": 4551.333333333333,
              "p95": 6000,
              "max": 6000
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 28,
          "strict_passes": 12,
          "checks_earned": 190,
          "checks_possible": 236,
          "check_percent": 80.508,
          "wall_s": 1629.666,
          "median_decode_tps": 41.26,
          "median_prefill_tps": 2574.097,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 1598.77,
              "min": 37.65,
              "median": 41.26,
              "mean": 40.99410256410256,
              "p95": 41.56,
              "max": 41.57
            },
            "prefill_tps": {
              "n": 39,
              "total": 100117.28547683333,
              "min": 2276.023540298243,
              "median": 2574.097130839442,
              "mean": 2567.109884021367,
              "p95": 2665.5645196563987,
              "max": 2674.246670284328
            },
            "elapsed_s": {
              "n": 39,
              "total": 1629.666,
              "min": 3.071,
              "median": 34.174,
              "mean": 41.786307692307695,
              "p95": 110.357,
              "max": 110.99
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 63690,
              "min": 64,
              "median": 1378,
              "mean": 1633.076923076923,
              "p95": 4098,
              "max": 4455
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 25,
          "strict_passes": 15,
          "checks_earned": 229,
          "checks_possible": 256,
          "check_percent": 89.453,
          "wall_s": 156.309,
          "median_decode_tps": 40.75,
          "median_prefill_tps": 1795.423,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1183.17,
              "min": 39.42,
              "median": 40.75,
              "mean": 40.79896551724138,
              "p95": 41.75,
              "max": 41.81
            },
            "prefill_tps": {
              "n": 29,
              "total": 51714.856586307906,
              "min": 999.3337774816789,
              "median": 1795.4226808283152,
              "mean": 1783.270916769238,
              "p95": 2580.461710711049,
              "max": 2626.5836577456944
            },
            "elapsed_s": {
              "n": 29,
              "total": 156.309,
              "min": 0.873,
              "median": 3.781,
              "mean": 5.38996551724138,
              "p95": 21.63,
              "max": 26.412
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 5766,
              "min": 8,
              "median": 131,
              "mean": 198.82758620689654,
              "p95": 859,
              "max": 1047
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-83",
      "run": "20260917T180723Z/akm-model-eval-moe-1x4060-131k-q8-ncpu40.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 88.163,
      "structure": 115,
      "strict": 77,
      "prefill": 106.08113495601683,
      "decode": 17.45,
      "ttft": null,
      "latency": 15.934,
      "wall": 7319.673,
      "requests": 125,
      "retries": 0,
      "truncated": 1,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260917T180723Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Explicit single-GPU run artifact name.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 115,
        "strict_passes": 77,
        "checks_earned": 648,
        "checks_possible": 735,
        "check_percent": 88.163,
        "wall_s": 7319.673,
        "median_decode_tps": 17.45,
        "median_prefill_tps": 106.081,
        "retries": 0,
        "truncated": 1,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 2129.5,
            "min": 13.16,
            "median": 17.45,
            "mean": 17.3130081300813,
            "p95": 17.87,
            "max": 18.12
          },
          "prefill_tps": {
            "n": 123,
            "total": 12844.179503491087,
            "min": 68.66082612100158,
            "median": 106.08113495601683,
            "mean": 104.4242236056186,
            "p95": 128.9592900034382,
            "max": 133.643280551968
          },
          "elapsed_s": {
            "n": 123,
            "total": 7319.673,
            "min": 2.682,
            "median": 15.934,
            "mean": 59.50953658536586,
            "p95": 231.072,
            "max": 863.72
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 77688,
            "min": 16,
            "median": 182,
            "mean": 631.609756097561,
            "p95": 2623,
            "max": 6000
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 43,
          "checks_earned": 202,
          "checks_possible": 213,
          "check_percent": 94.836,
          "wall_s": 532.622,
          "median_decode_tps": 17.425,
          "median_prefill_tps": 98.39,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 905.1,
              "min": 16.73,
              "median": 17.425,
              "mean": 17.40576923076923,
              "p95": 17.91,
              "max": 17.98
            },
            "prefill_tps": {
              "n": 52,
              "total": 5022.141635461814,
              "min": 71.39736279871452,
              "median": 98.3899611148682,
              "mean": 96.57964683580411,
              "p95": 119.21557315939877,
              "max": 120.19407663966798
            },
            "elapsed_s": {
              "n": 52,
              "total": 532.622,
              "min": 2.682,
              "median": 7.579,
              "mean": 10.24273076923077,
              "p95": 26.304,
              "max": 36.221
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 6491,
              "min": 25,
              "median": 75,
              "mean": 124.82692307692308,
              "p95": 404,
              "max": 566
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 2,
          "strict_passes": 0,
          "checks_earned": 23,
          "checks_possible": 30,
          "check_percent": 76.667,
          "wall_s": 1407.612,
          "median_decode_tps": 14.51,
          "median_prefill_tps": 127.331,
          "retries": 0,
          "truncated": 1,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 42.95,
              "min": 13.16,
              "median": 14.51,
              "mean": 14.316666666666666,
              "p95": 15.28,
              "max": 15.28
            },
            "prefill_tps": {
              "n": 3,
              "total": 377.4851193866672,
              "min": 122.73721440590565,
              "median": 127.3307484621347,
              "mean": 125.82837312888907,
              "p95": 127.41715651862685,
              "max": 127.41715651862685
            },
            "elapsed_s": {
              "n": 3,
              "total": 1407.612,
              "min": 231.072,
              "median": 312.82,
              "mean": 469.204,
              "p95": 863.72,
              "max": 863.72
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 6819,
              "min": 374,
              "median": 445,
              "mean": 2273,
              "p95": 6000,
              "max": 6000
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 35,
          "strict_passes": 15,
          "checks_earned": 193,
          "checks_possible": 236,
          "check_percent": 81.78,
          "wall_s": 4732.257,
          "median_decode_tps": 17.43,
          "median_prefill_tps": 112.142,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 673.74,
              "min": 15,
              "median": 17.43,
              "mean": 17.275384615384617,
              "p95": 17.7,
              "max": 17.71
            },
            "prefill_tps": {
              "n": 39,
              "total": 4465.266506781363,
              "min": 103.23702983501303,
              "median": 112.14214621079688,
              "mean": 114.49401299439394,
              "p95": 132.93454491024522,
              "max": 133.643280551968
            },
            "elapsed_s": {
              "n": 39,
              "total": 4732.257,
              "min": 27.207,
              "median": 93.819,
              "mean": 121.33992307692309,
              "p95": 300.555,
              "max": 437.286
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 57287,
              "min": 46,
              "median": 1240,
              "mean": 1468.8974358974358,
              "p95": 3308,
              "max": 3996
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 26,
          "strict_passes": 19,
          "checks_earned": 230,
          "checks_possible": 256,
          "check_percent": 89.844,
          "wall_s": 647.182,
          "median_decode_tps": 17.5,
          "median_prefill_tps": 105.127,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 507.71,
              "min": 16.44,
              "median": 17.5,
              "mean": 17.507241379310344,
              "p95": 17.95,
              "max": 18.12
            },
            "prefill_tps": {
              "n": 29,
              "total": 2979.2862418612426,
              "min": 68.66082612100158,
              "median": 105.12688501729546,
              "mean": 102.73400834004285,
              "p95": 128.9592900034382,
              "max": 133.62189205143991
            },
            "elapsed_s": {
              "n": 29,
              "total": 647.182,
              "min": 2.945,
              "median": 14.491,
              "mean": 22.31662068965517,
              "p95": 83.137,
              "max": 136.286
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 7091,
              "min": 16,
              "median": 142,
              "mean": 244.51724137931035,
              "p95": 937,
              "max": 2220
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-84",
      "run": "20260917T180723Z/cold-cache-probe.jsonl",
      "model": "Flash Next · Q2_K_XL",
      "model_family": "Flash Next",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "partial/probe",
      "cases": 1,
      "expected": 123,
      "success": 100,
      "quality": 100,
      "structure": 1,
      "strict": 1,
      "prefill": 9.518755189618432,
      "decode": 7.55,
      "ttft": null,
      "latency": 97.323,
      "wall": 97.323,
      "requests": 1,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "20260917T180723Z",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": 1,
        "checks_earned": 10,
        "checks_possible": 10,
        "check_percent": 100,
        "wall_s": 97.323,
        "median_decode_tps": 7.55,
        "median_prefill_tps": 9.519,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 1,
        "request_count_recorded_cases": 1,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": 1,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 7.55,
            "min": 7.55,
            "median": 7.55,
            "mean": 7.55,
            "p95": 7.55,
            "max": 7.55
          },
          "prefill_tps": {
            "n": 1,
            "total": 9.518755189618432,
            "min": 9.518755189618432,
            "median": 9.518755189618432,
            "mean": 9.518755189618432,
            "p95": 9.518755189618432,
            "max": 9.518755189618432
          },
          "elapsed_s": {
            "n": 1,
            "total": 97.323,
            "min": 97.323,
            "median": 97.323,
            "mean": 97.323,
            "p95": 97.323,
            "max": 97.323
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 1,
            "total": 621,
            "min": 621,
            "median": 621,
            "mean": 621,
            "p95": 621,
            "max": 621
          },
          "completion_tokens": {
            "n": 1,
            "total": 243,
            "min": 243,
            "median": 243,
            "mean": 243,
            "p95": 243,
            "max": 243
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": 1,
          "checks_earned": 10,
          "checks_possible": 10,
          "check_percent": 100,
          "wall_s": 97.323,
          "median_decode_tps": 7.55,
          "median_prefill_tps": 9.519,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 1,
          "request_count_recorded_cases": 1,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": 1,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 7.55,
              "min": 7.55,
              "median": 7.55,
              "mean": 7.55,
              "p95": 7.55,
              "max": 7.55
            },
            "prefill_tps": {
              "n": 1,
              "total": 9.518755189618432,
              "min": 9.518755189618432,
              "median": 9.518755189618432,
              "mean": 9.518755189618432,
              "p95": 9.518755189618432,
              "max": 9.518755189618432
            },
            "elapsed_s": {
              "n": 1,
              "total": 97.323,
              "min": 97.323,
              "median": 97.323,
              "mean": 97.323,
              "p95": 97.323,
              "max": 97.323
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 621,
              "min": 621,
              "median": 621,
              "mean": 621,
              "p95": 621,
              "max": 621
            },
            "completion_tokens": {
              "n": 1,
              "total": 243,
              "min": 243,
              "median": 243,
              "mean": 243,
              "p95": 243,
              "max": 243
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-85",
      "run": "20260917T234720Z/akm-model-eval-qwen36-35b-131k-q8.jsonl",
      "model": "Qwen3.6-35B-A3B · IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 85.034,
      "structure": 101,
      "strict": 63,
      "prefill": 1575.12522066105,
      "decode": 61.45,
      "ttft": null,
      "latency": 3.672,
      "wall": 1501.458,
      "requests": 125,
      "retries": 0,
      "truncated": 1,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260917T234720Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies a two-GPU endpoint.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 101,
        "strict_passes": 63,
        "checks_earned": 625,
        "checks_possible": 735,
        "check_percent": 85.034,
        "wall_s": 1501.458,
        "median_decode_tps": 61.45,
        "median_prefill_tps": 1575.125,
        "retries": 0,
        "truncated": 1,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 7539.51,
            "min": 48.42,
            "median": 61.45,
            "mean": 61.29682926829268,
            "p95": 63.2,
            "max": 63.9
          },
          "prefill_tps": {
            "n": 123,
            "total": 233591.1524797241,
            "min": 519.2401033965076,
            "median": 1575.12522066105,
            "mean": 1899.1150608107653,
            "p95": 3928.5076948984697,
            "max": 4216.005752825721
          },
          "elapsed_s": {
            "n": 123,
            "total": 1501.458,
            "min": 0.711,
            "median": 3.672,
            "mean": 12.206975609756098,
            "p95": 41.1,
            "max": 108.169
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 84394,
            "min": 16,
            "median": 190,
            "mean": 686.130081300813,
            "p95": 2516,
            "max": 6000
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 36,
          "checks_earned": 190,
          "checks_possible": 213,
          "check_percent": 89.202,
          "wall_s": 140.579,
          "median_decode_tps": 61.24,
          "median_prefill_tps": 1219.582,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 3193.26,
              "min": 60.43,
              "median": 61.239999999999995,
              "mean": 61.408846153846156,
              "p95": 63.28,
              "max": 63.66
            },
            "prefill_tps": {
              "n": 52,
              "total": 61671.205661930624,
              "min": 519.2401033965076,
              "median": 1219.582456247571,
              "mean": 1185.9847242678966,
              "p95": 1839.3816088756892,
              "max": 2260.6832428847465
            },
            "elapsed_s": {
              "n": 52,
              "total": 140.579,
              "min": 1.126,
              "median": 1.7745000000000002,
              "mean": 2.7034423076923075,
              "p95": 6.714,
              "max": 14.079
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7641,
              "min": 46,
              "median": 89.5,
              "mean": 146.94230769230768,
              "p95": 385,
              "max": 869
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 2,
          "strict_passes": 1,
          "checks_earned": 24,
          "checks_possible": 30,
          "check_percent": 80,
          "wall_s": 61.365,
          "median_decode_tps": 52.58,
          "median_prefill_tps": 3928.508,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 156.06,
              "min": 48.42,
              "median": 52.58,
              "mean": 52.02,
              "p95": 55.06,
              "max": 55.06
            },
            "prefill_tps": {
              "n": 3,
              "total": 11698.281853698618,
              "min": 3658.2963732633893,
              "median": 3928.5076948984697,
              "mean": 3899.4272845662063,
              "p95": 4111.477785536759,
              "max": 4111.477785536759
            },
            "elapsed_s": {
              "n": 3,
              "total": 61.365,
              "min": 14.205,
              "median": 17.692,
              "mean": 20.455,
              "p95": 29.468,
              "max": 29.468
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 1609,
              "min": 421,
              "median": 445,
              "mean": 536.3333333333334,
              "p95": 743,
              "max": 743
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 21,
          "strict_passes": 12,
          "checks_earned": 180,
          "checks_possible": 236,
          "check_percent": 76.271,
          "wall_s": 1131.669,
          "median_decode_tps": 62.33,
          "median_prefill_tps": 2943.449,
          "retries": 0,
          "truncated": 1,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 2406.71,
              "min": 53.91,
              "median": 62.33,
              "mean": 61.71051282051282,
              "p95": 63.11,
              "max": 63.2
            },
            "prefill_tps": {
              "n": 39,
              "total": 115888.84066793213,
              "min": 2259.030407387372,
              "median": 2943.4491733940663,
              "mean": 2971.508735075183,
              "p95": 4095.224740080693,
              "max": 4216.005752825721
            },
            "elapsed_s": {
              "n": 39,
              "total": 1131.669,
              "min": 2.533,
              "median": 24.98,
              "mean": 29.017153846153846,
              "p95": 65.441,
              "max": 108.169
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 65845,
              "min": 48,
              "median": 1498,
              "mean": 1688.3333333333333,
              "p95": 3892,
              "max": 6000
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 26,
          "strict_passes": 14,
          "checks_earned": 231,
          "checks_possible": 256,
          "check_percent": 90.234,
          "wall_s": 167.845,
          "median_decode_tps": 61.26,
          "median_prefill_tps": 1384.767,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1783.48,
              "min": 59.61,
              "median": 61.26,
              "mean": 61.499310344827585,
              "p95": 63.46,
              "max": 63.9
            },
            "prefill_tps": {
              "n": 29,
              "total": 44332.82429616275,
              "min": 668.9780033661274,
              "median": 1384.7667873799512,
              "mean": 1528.7180791780258,
              "p95": 3172.2284074364725,
              "max": 4006.1386948625745
            },
            "elapsed_s": {
              "n": 29,
              "total": 167.845,
              "min": 0.711,
              "median": 2.97,
              "mean": 5.787758620689655,
              "p95": 25.457,
              "max": 36.136
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 9299,
              "min": 16,
              "median": 156,
              "mean": 320.6551724137931,
              "p95": 1533,
              "max": 1998
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-86",
      "run": "20260917T234720Z/smoke-probe-excluded.jsonl",
      "model": "Qwen3.6-35B-A3B · IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "partial/probe",
      "cases": 1,
      "expected": 123,
      "success": 100,
      "quality": 70,
      "structure": 1,
      "strict": 0,
      "prefill": 932.6536926177682,
      "decode": 61.98,
      "ttft": null,
      "latency": 2.91,
      "wall": 2.91,
      "requests": 1,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "20260917T234720Z",
      "duplicate_of": null,
      "unique_cases": 1,
      "gpu_count": 2,
      "gpu_used": "2 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies a two-GPU endpoint.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 1,
        "ok": 1,
        "structure": 1,
        "strict_passes": 0,
        "checks_earned": 7,
        "checks_possible": 10,
        "check_percent": 70,
        "wall_s": 2.91,
        "median_decode_tps": 61.98,
        "median_prefill_tps": 932.654,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 1,
        "retry_count_recorded_cases": 1,
        "request_count_recorded_cases": 1,
        "finish_reason_recorded_cases": 1,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 1,
        "http_requests": 1,
        "metrics": {
          "decode_tps": {
            "n": 1,
            "total": 61.98,
            "min": 61.98,
            "median": 61.98,
            "mean": 61.98,
            "p95": 61.98,
            "max": 61.98
          },
          "prefill_tps": {
            "n": 1,
            "total": 932.6536926177682,
            "min": 932.6536926177682,
            "median": 932.6536926177682,
            "mean": 932.6536926177682,
            "p95": 932.6536926177682,
            "max": 932.6536926177682
          },
          "elapsed_s": {
            "n": 1,
            "total": 2.91,
            "min": 2.91,
            "median": 2.91,
            "mean": 2.91,
            "p95": 2.91,
            "max": 2.91
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 1,
            "total": 621,
            "min": 621,
            "median": 621,
            "mean": 621,
            "p95": 621,
            "max": 621
          },
          "completion_tokens": {
            "n": 1,
            "total": 139,
            "min": 139,
            "median": 139,
            "mean": 139,
            "p95": 139,
            "max": 139
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 1,
          "ok": 1,
          "structure": 1,
          "strict_passes": 0,
          "checks_earned": 7,
          "checks_possible": 10,
          "check_percent": 70,
          "wall_s": 2.91,
          "median_decode_tps": 61.98,
          "median_prefill_tps": 932.654,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 1,
          "retry_count_recorded_cases": 1,
          "request_count_recorded_cases": 1,
          "finish_reason_recorded_cases": 1,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 1,
          "http_requests": 1,
          "metrics": {
            "decode_tps": {
              "n": 1,
              "total": 61.98,
              "min": 61.98,
              "median": 61.98,
              "mean": 61.98,
              "p95": 61.98,
              "max": 61.98
            },
            "prefill_tps": {
              "n": 1,
              "total": 932.6536926177682,
              "min": 932.6536926177682,
              "median": 932.6536926177682,
              "mean": 932.6536926177682,
              "p95": 932.6536926177682,
              "max": 932.6536926177682
            },
            "elapsed_s": {
              "n": 1,
              "total": 2.91,
              "min": 2.91,
              "median": 2.91,
              "mean": 2.91,
              "p95": 2.91,
              "max": 2.91
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 1,
              "total": 621,
              "min": 621,
              "median": 621,
              "mean": 621,
              "p95": 621,
              "max": 621
            },
            "completion_tokens": {
              "n": 1,
              "total": 139,
              "min": 139,
              "median": 139,
              "mean": 139,
              "p95": 139,
              "max": 139
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-87",
      "run": "20260918T000000Z/akm-model-eval-qwen36-35b-b70-131k-q8.jsonl",
      "model": "Qwen3.6-35B-A3B · IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "llama.cpp SYCL",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 85.442,
      "structure": 105,
      "strict": 63,
      "prefill": 524.5549892506027,
      "decode": 67.61,
      "ttft": null,
      "latency": 4.348,
      "wall": 1706.142,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260918T000000Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 105,
        "strict_passes": 63,
        "checks_earned": 628,
        "checks_possible": 735,
        "check_percent": 85.442,
        "wall_s": 1706.142,
        "median_decode_tps": 67.61,
        "median_prefill_tps": 524.555,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 8090.48,
            "min": 45.42,
            "median": 67.61,
            "mean": 65.77626016260163,
            "p95": 69.66,
            "max": 70.44
          },
          "prefill_tps": {
            "n": 123,
            "total": 63766.05776084735,
            "min": 171.85969374602575,
            "median": 524.5549892506027,
            "mean": 518.4232338280272,
            "p95": 769.2384306434075,
            "max": 799.8961349340142
          },
          "elapsed_s": {
            "n": 123,
            "total": 1706.142,
            "min": 1.014,
            "median": 4.348,
            "mean": 13.871073170731707,
            "p95": 57.391,
            "max": 95.756
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 78515,
            "min": 16,
            "median": 193,
            "mean": 638.3333333333334,
            "p95": 2448,
            "max": 4006
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 52,
          "strict_passes": 36,
          "checks_earned": 192,
          "checks_possible": 213,
          "check_percent": 90.141,
          "wall_s": 149.054,
          "median_decode_tps": 68.395,
          "median_prefill_tps": 394.356,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 3475.33,
              "min": 50.36,
              "median": 68.39500000000001,
              "mean": 66.83326923076923,
              "p95": 69.66,
              "max": 69.96
            },
            "prefill_tps": {
              "n": 52,
              "total": 20692.614787777733,
              "min": 171.85969374602575,
              "median": 394.35614635743,
              "mean": 397.9348997649564,
              "p95": 626.2391133676085,
              "max": 727.5424553705279
            },
            "elapsed_s": {
              "n": 52,
              "total": 149.054,
              "min": 1.357,
              "median": 2.1055,
              "mean": 2.866423076923077,
              "p95": 6.717,
              "max": 8.996
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7183,
              "min": 46,
              "median": 91.5,
              "mean": 138.1346153846154,
              "p95": 387,
              "max": 566
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 0,
          "checks_earned": 27,
          "checks_possible": 30,
          "check_percent": 90,
          "wall_s": 172.365,
          "median_decode_tps": 50.35,
          "median_prefill_tps": 788.367,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 150.2,
              "min": 45.42,
              "median": 50.35,
              "mean": 50.06666666666667,
              "p95": 54.43,
              "max": 54.43
            },
            "prefill_tps": {
              "n": 3,
              "total": 2364.575249116456,
              "min": 776.312475032254,
              "median": 788.3666391501877,
              "mean": 788.1917497054853,
              "p95": 799.8961349340142,
              "max": 799.8961349340142
            },
            "elapsed_s": {
              "n": 3,
              "total": 172.365,
              "min": 41.692,
              "median": 57.391,
              "mean": 57.455,
              "p95": 73.282,
              "max": 73.282
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 1454,
              "min": 389,
              "median": 474,
              "mean": 484.6666666666667,
              "p95": 591,
              "max": 591
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 23,
          "strict_passes": 13,
          "checks_earned": 182,
          "checks_possible": 236,
          "check_percent": 77.119,
          "wall_s": 1191.854,
          "median_decode_tps": 65.22,
          "median_prefill_tps": 709.014,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 2486.33,
              "min": 48.28,
              "median": 65.22,
              "mean": 63.75205128205128,
              "p95": 68.51,
              "max": 69.18
            },
            "prefill_tps": {
              "n": 39,
              "total": 27252.81118127636,
              "min": 524.5549892506027,
              "median": 709.0139843392817,
              "mean": 698.7900302891375,
              "p95": 778.6282898567491,
              "max": 792.0443825562671
            },
            "elapsed_s": {
              "n": 39,
              "total": 1191.854,
              "min": 5.905,
              "median": 23.471,
              "mean": 30.560358974358973,
              "p95": 93.399,
              "max": 95.756
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 60359,
              "min": 46,
              "median": 1375,
              "mean": 1547.6666666666667,
              "p95": 3669,
              "max": 4006
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 27,
          "strict_passes": 14,
          "checks_earned": 227,
          "checks_possible": 256,
          "check_percent": 88.672,
          "wall_s": 192.869,
          "median_decode_tps": 68.43,
          "median_prefill_tps": 462.129,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 1978.62,
              "min": 61.7,
              "median": 68.43,
              "mean": 68.22827586206897,
              "p95": 69.89,
              "max": 70.44
            },
            "prefill_tps": {
              "n": 29,
              "total": 13456.056542676795,
              "min": 224.57491177414178,
              "median": 462.12907264999114,
              "mean": 464.0019497474757,
              "p95": 769.2384306434075,
              "max": 794.4823793222677
            },
            "elapsed_s": {
              "n": 29,
              "total": 192.869,
              "min": 1.014,
              "median": 3.763,
              "mean": 6.650655172413793,
              "p95": 24.509,
              "max": 42.396
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 9519,
              "min": 16,
              "median": 189,
              "mean": 328.2413793103448,
              "p95": 1473,
              "max": 1843
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-88",
      "run": "20260918T003020Z/akm-model-eval-qwen36-35b-1gpu-cpu40-131k-q8.jsonl",
      "model": "Qwen3.6-35B-A3B · IQ4_XS",
      "model_family": "35B MoE",
      "quant": "IQ4_XS",
      "runtime": "llama.cpp CUDA",
      "suite": "public123",
      "status": "complete",
      "cases": 123,
      "expected": 123,
      "success": 100,
      "quality": 86.395,
      "structure": 101,
      "strict": 63,
      "prefill": 242.27076772975298,
      "decode": 17.3,
      "ttft": null,
      "latency": 14.195,
      "wall": 5758.366,
      "requests": 125,
      "retries": 0,
      "truncated": 0,
      "cache": null,
      "context": 131072,
      "slots": null,
      "kv": null,
      "date": "20260918T003020Z",
      "duplicate_of": null,
      "unique_cases": 123,
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Explicit single-GPU run artifact name.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 123,
        "ok": 123,
        "structure": 101,
        "strict_passes": 63,
        "checks_earned": 635,
        "checks_possible": 735,
        "check_percent": 86.395,
        "wall_s": 5758.366,
        "median_decode_tps": 17.3,
        "median_prefill_tps": 242.271,
        "retries": 0,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 123,
        "retry_count_recorded_cases": 123,
        "request_count_recorded_cases": 123,
        "finish_reason_recorded_cases": 123,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 123,
        "http_requests": 125,
        "metrics": {
          "decode_tps": {
            "n": 123,
            "total": 2092.51,
            "min": 4.45,
            "median": 17.3,
            "mean": 17.012276422764227,
            "p95": 26.96,
            "max": 27.23
          },
          "prefill_tps": {
            "n": 123,
            "total": 30342.483426383147,
            "min": 101.27462927431223,
            "median": 242.27076772975298,
            "mean": 246.68685712506624,
            "p95": 350.80707985092835,
            "max": 355.6538941900708
          },
          "elapsed_s": {
            "n": 123,
            "total": 5758.366,
            "min": 2.545,
            "median": 14.195,
            "mean": 46.8159837398374,
            "p95": 177.4,
            "max": 363.174
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 123,
            "total": 317145,
            "min": 92,
            "median": 456,
            "mean": 2578.4146341463415,
            "p95": 9823,
            "max": 49960
          },
          "completion_tokens": {
            "n": 123,
            "total": 77743,
            "min": 16,
            "median": 200,
            "mean": 632.0569105691056,
            "p95": 2437,
            "max": 3787
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "compact/focused": {
          "cases": 52,
          "ok": 52,
          "structure": 51,
          "strict_passes": 38,
          "checks_earned": 197,
          "checks_possible": 213,
          "check_percent": 92.488,
          "wall_s": 522.914,
          "median_decode_tps": 17.465,
          "median_prefill_tps": 199.097,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 52,
          "retry_count_recorded_cases": 52,
          "request_count_recorded_cases": 52,
          "finish_reason_recorded_cases": 52,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 52,
          "http_requests": 52,
          "metrics": {
            "decode_tps": {
              "n": 52,
              "total": 906.21,
              "min": 5.17,
              "median": 17.465,
              "mean": 17.427115384615384,
              "p95": 27.21,
              "max": 27.23
            },
            "prefill_tps": {
              "n": 52,
              "total": 10170.74236991339,
              "min": 101.27462927431223,
              "median": 199.09717906606556,
              "mean": 195.59119942141135,
              "p95": 269.01936592871897,
              "max": 302.8512872034011
            },
            "elapsed_s": {
              "n": 52,
              "total": 522.914,
              "min": 3.742,
              "median": 9.185500000000001,
              "mean": 10.056038461538462,
              "p95": 25.01,
              "max": 35.552
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 52,
              "total": 15883,
              "min": 92,
              "median": 293.5,
              "mean": 305.4423076923077,
              "p95": 564,
              "max": 709
            },
            "completion_tokens": {
              "n": 52,
              "total": 7499,
              "min": 46,
              "median": 95.5,
              "mean": 144.21153846153845,
              "p95": 385,
              "max": 596
            },
            "cached_prompt_tokens": null
          }
        },
        "context/context": {
          "cases": 3,
          "ok": 3,
          "structure": 3,
          "strict_passes": 1,
          "checks_earned": 27,
          "checks_possible": 30,
          "check_percent": 90,
          "wall_s": 429.41600000000005,
          "median_decode_tps": 17.61,
          "median_prefill_tps": 341.174,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 3,
          "retry_count_recorded_cases": 3,
          "request_count_recorded_cases": 3,
          "finish_reason_recorded_cases": 3,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 3,
          "http_requests": 3,
          "metrics": {
            "decode_tps": {
              "n": 3,
              "total": 43.989999999999995,
              "min": 8.29,
              "median": 17.61,
              "mean": 14.663333333333332,
              "p95": 18.09,
              "max": 18.09
            },
            "prefill_tps": {
              "n": 3,
              "total": 1011.2667552638912,
              "min": 319.28597124737803,
              "median": 341.17370416558487,
              "mean": 337.0889184212971,
              "p95": 350.80707985092835,
              "max": 350.80707985092835
            },
            "elapsed_s": {
              "n": 3,
              "total": 429.41600000000005,
              "min": 102.927,
              "median": 119.558,
              "mean": 143.13866666666667,
              "p95": 206.931,
              "max": 206.931
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 3,
              "total": 112192,
              "min": 26283,
              "median": 35949,
              "mean": 37397.333333333336,
              "p95": 49960,
              "max": 49960
            },
            "completion_tokens": {
              "n": 3,
              "total": 1167,
              "min": 249,
              "median": 414,
              "mean": 389,
              "p95": 504,
              "max": 504
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/legacy": {
          "cases": 39,
          "ok": 39,
          "structure": 21,
          "strict_passes": 10,
          "checks_earned": 183,
          "checks_possible": 236,
          "check_percent": 77.542,
          "wall_s": 4144.137,
          "median_decode_tps": 17.49,
          "median_prefill_tps": 332.684,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 39,
          "retry_count_recorded_cases": 39,
          "request_count_recorded_cases": 39,
          "finish_reason_recorded_cases": 39,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 39,
          "http_requests": 39,
          "metrics": {
            "decode_tps": {
              "n": 39,
              "total": 645.94,
              "min": 4.45,
              "median": 17.49,
              "mean": 16.562564102564103,
              "p95": 19,
              "max": 19.3
            },
            "prefill_tps": {
              "n": 39,
              "total": 13046.732703330537,
              "min": 309.1941997670876,
              "median": 332.68437229722497,
              "mean": 334.5316077777061,
              "p95": 352.15299737482684,
              "max": 355.6538941900708
            },
            "elapsed_s": {
              "n": 39,
              "total": 4144.137,
              "min": 12.235,
              "median": 94.789,
              "mean": 106.25992307692307,
              "p95": 241.326,
              "max": 363.174
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 39,
              "total": 160623,
              "min": 1186,
              "median": 2862,
              "mean": 4118.538461538462,
              "p95": 15220,
              "max": 29796
            },
            "completion_tokens": {
              "n": 39,
              "total": 60120,
              "min": 46,
              "median": 1494,
              "mean": 1541.5384615384614,
              "p95": 3361,
              "max": 3787
            },
            "cached_prompt_tokens": null
          }
        },
        "deep/production": {
          "cases": 29,
          "ok": 29,
          "structure": 26,
          "strict_passes": 14,
          "checks_earned": 228,
          "checks_possible": 256,
          "check_percent": 89.062,
          "wall_s": 661.899,
          "median_decode_tps": 16.66,
          "median_prefill_tps": 204.251,
          "retries": 0,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 29,
          "retry_count_recorded_cases": 29,
          "request_count_recorded_cases": 29,
          "finish_reason_recorded_cases": 29,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 29,
          "http_requests": 31,
          "metrics": {
            "decode_tps": {
              "n": 29,
              "total": 496.37,
              "min": 7.23,
              "median": 16.66,
              "mean": 17.116206896551724,
              "p95": 19.27,
              "max": 19.39
            },
            "prefill_tps": {
              "n": 29,
              "total": 6113.741597875328,
              "min": 121.850974324259,
              "median": 204.25065317656797,
              "mean": 210.81867578880443,
              "p95": 344.33272444304174,
              "max": 354.1279337361492
            },
            "elapsed_s": {
              "n": 29,
              "total": 661.899,
              "min": 2.545,
              "median": 12.785,
              "mean": 22.824103448275864,
              "p95": 106.251,
              "max": 110.657
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 29,
              "total": 28447,
              "min": 126,
              "median": 371,
              "mean": 980.9310344827586,
              "p95": 3630,
              "max": 9823
            },
            "completion_tokens": {
              "n": 29,
              "total": 8957,
              "min": 16,
              "median": 154,
              "mean": 308.86206896551727,
              "p95": 1386,
              "max": 1472
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-89",
      "run": "raw-completion/flag-raw-nprobs10.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "raw-completion diagnostic",
      "status": "diagnostic",
      "cases": 2,
      "expected": null,
      "success": 100,
      "quality": null,
      "structure": null,
      "strict": null,
      "prefill": 22.474324980187074,
      "decode": 39.679333809896846,
      "ttft": null,
      "latency": 32.15,
      "wall": 64.3,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 13,
        "checks_possible": 13,
        "check_percent": 100,
        "wall_s": 64.3,
        "median_decode_tps": 39.679,
        "median_prefill_tps": 22.474,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 79.35866761979369,
            "min": 39.15072303789042,
            "median": 39.679333809896846,
            "mean": 39.679333809896846,
            "p95": 40.20794458190328,
            "max": 40.20794458190328
          },
          "prefill_tps": {
            "n": 2,
            "total": 44.94864996037415,
            "min": 22.27059891208124,
            "median": 22.474324980187074,
            "mean": 22.474324980187074,
            "p95": 22.678051048292907,
            "max": 22.678051048292907
          },
          "elapsed_s": {
            "n": 2,
            "total": 64.3,
            "min": 26.3,
            "median": 32.15,
            "mean": 32.15,
            "p95": 38,
            "max": 38
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 6151,
            "min": 2342,
            "median": 3075.5,
            "mean": 3075.5,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 2,
            "total": 2511,
            "min": 1006,
            "median": 1255.5,
            "mean": 1255.5,
            "p95": 1505,
            "max": 1505
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 2,
          "ok": 2,
          "structure": 2,
          "strict_passes": null,
          "checks_earned": 13,
          "checks_possible": 13,
          "check_percent": 100,
          "wall_s": 64.3,
          "median_decode_tps": 39.679,
          "median_prefill_tps": 22.474,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 2,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 2,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 2,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 2,
              "total": 79.35866761979369,
              "min": 39.15072303789042,
              "median": 39.679333809896846,
              "mean": 39.679333809896846,
              "p95": 40.20794458190328,
              "max": 40.20794458190328
            },
            "prefill_tps": {
              "n": 2,
              "total": 44.94864996037415,
              "min": 22.27059891208124,
              "median": 22.474324980187074,
              "mean": 22.474324980187074,
              "p95": 22.678051048292907,
              "max": 22.678051048292907
            },
            "elapsed_s": {
              "n": 2,
              "total": 64.3,
              "min": 26.3,
              "median": 32.15,
              "mean": 32.15,
              "p95": 38,
              "max": 38
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 2,
              "total": 6151,
              "min": 2342,
              "median": 3075.5,
              "mean": 3075.5,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 2,
              "total": 2511,
              "min": 1006,
              "median": 1255.5,
              "mean": 1255.5,
              "p95": 1505,
              "max": 1505
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-90",
      "run": "raw-completion/flag-raw-tokens.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA",
      "suite": "raw-completion diagnostic",
      "status": "diagnostic",
      "cases": 5,
      "expected": null,
      "success": 100,
      "quality": null,
      "structure": null,
      "strict": null,
      "prefill": 22.527723179337574,
      "decode": 39.69329970546115,
      "ttft": null,
      "latency": 30.8,
      "wall": 163.1,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 25,
        "checks_possible": 32,
        "check_percent": 78.125,
        "wall_s": 163.1,
        "median_decode_tps": 39.693,
        "median_prefill_tps": 22.528,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 199.9066201546628,
            "min": 39.47496897444156,
            "median": 39.69329970546115,
            "mean": 39.981324030932555,
            "p95": 40.632267147047976,
            "max": 40.632267147047976
          },
          "prefill_tps": {
            "n": 5,
            "total": 112.41591140044883,
            "min": 22.235934381757637,
            "median": 22.527723179337574,
            "mean": 22.483182280089768,
            "p95": 22.686282739141777,
            "max": 22.686282739141777
          },
          "elapsed_s": {
            "n": 5,
            "total": 163.1,
            "min": 25.9,
            "median": 30.8,
            "mean": 32.620000000000005,
            "p95": 40.7,
            "max": 40.7
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6420,
            "min": 1006,
            "median": 1231,
            "mean": 1284,
            "p95": 1587,
            "max": 1587
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 25,
          "checks_possible": 32,
          "check_percent": 78.125,
          "wall_s": 163.1,
          "median_decode_tps": 39.693,
          "median_prefill_tps": 22.528,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 199.9066201546628,
              "min": 39.47496897444156,
              "median": 39.69329970546115,
              "mean": 39.981324030932555,
              "p95": 40.632267147047976,
              "max": 40.632267147047976
            },
            "prefill_tps": {
              "n": 5,
              "total": 112.41591140044883,
              "min": 22.235934381757637,
              "median": 22.527723179337574,
              "mean": 22.483182280089768,
              "p95": 22.686282739141777,
              "max": 22.686282739141777
            },
            "elapsed_s": {
              "n": 5,
              "total": 163.1,
              "min": 25.9,
              "median": 30.8,
              "mean": 32.620000000000005,
              "p95": 40.7,
              "max": 40.7
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6420,
              "min": 1006,
              "median": 1231,
              "mean": 1284,
              "p95": 1587,
              "max": 1587
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-91",
      "run": "raw-completion/lms-raw-nprobs10.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "raw-completion diagnostic",
      "status": "diagnostic",
      "cases": 2,
      "expected": null,
      "success": 100,
      "quality": null,
      "structure": null,
      "strict": null,
      "prefill": 22.087713054288184,
      "decode": 47.19512721388952,
      "ttft": null,
      "latency": 26.4,
      "wall": 52.8,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 2,
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 2,
        "ok": 2,
        "structure": 2,
        "strict_passes": null,
        "checks_earned": 13,
        "checks_possible": 13,
        "check_percent": 100,
        "wall_s": 52.8,
        "median_decode_tps": 47.195,
        "median_prefill_tps": 22.088,
        "retries": null,
        "truncated": 0,
        "near_quotes": 0,
        "fabricated_quotes": 0,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 2,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 2,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 2,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 2,
            "total": 94.39025442777904,
            "min": 46.16120008766314,
            "median": 47.19512721388952,
            "mean": 47.19512721388952,
            "p95": 48.22905434011591,
            "max": 48.22905434011591
          },
          "prefill_tps": {
            "n": 2,
            "total": 44.17542610857637,
            "min": 21.924896267834534,
            "median": 22.087713054288184,
            "mean": 22.087713054288184,
            "p95": 22.250529840741834,
            "max": 22.250529840741834
          },
          "elapsed_s": {
            "n": 2,
            "total": 52.8,
            "min": 23.9,
            "median": 26.4,
            "mean": 26.4,
            "p95": 28.9,
            "max": 28.9
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 2,
            "total": 6151,
            "min": 2342,
            "median": 3075.5,
            "mean": 3075.5,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 2,
            "total": 2430,
            "min": 1071,
            "median": 1215,
            "mean": 1215,
            "p95": 1359,
            "max": 1359
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 2,
          "ok": 2,
          "structure": 2,
          "strict_passes": null,
          "checks_earned": 13,
          "checks_possible": 13,
          "check_percent": 100,
          "wall_s": 52.8,
          "median_decode_tps": 47.195,
          "median_prefill_tps": 22.088,
          "retries": null,
          "truncated": 0,
          "near_quotes": 0,
          "fabricated_quotes": 0,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 2,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 2,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 2,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 2,
              "total": 94.39025442777904,
              "min": 46.16120008766314,
              "median": 47.19512721388952,
              "mean": 47.19512721388952,
              "p95": 48.22905434011591,
              "max": 48.22905434011591
            },
            "prefill_tps": {
              "n": 2,
              "total": 44.17542610857637,
              "min": 21.924896267834534,
              "median": 22.087713054288184,
              "mean": 22.087713054288184,
              "p95": 22.250529840741834,
              "max": 22.250529840741834
            },
            "elapsed_s": {
              "n": 2,
              "total": 52.8,
              "min": 23.9,
              "median": 26.4,
              "mean": 26.4,
              "p95": 28.9,
              "max": 28.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 2,
              "total": 6151,
              "min": 2342,
              "median": 3075.5,
              "mean": 3075.5,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 2,
              "total": 2430,
              "min": 1071,
              "median": 1215,
              "mean": 1215,
              "p95": 1359,
              "max": 1359
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-92",
      "run": "raw-completion/lms-raw-tokens.jsonl",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "LM Studio",
      "suite": "raw-completion diagnostic",
      "status": "diagnostic",
      "cases": 5,
      "expected": null,
      "success": 100,
      "quality": null,
      "structure": null,
      "strict": null,
      "prefill": 22.14361239820858,
      "decode": 46.53209754296622,
      "ttft": null,
      "latency": 25.8,
      "wall": 139.9,
      "requests": null,
      "retries": null,
      "truncated": 0,
      "cache": null,
      "context": null,
      "slots": null,
      "kv": null,
      "date": "historical · date not in filename",
      "duplicate_of": null,
      "unique_cases": 5,
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "cases": 5,
        "ok": 5,
        "structure": 4,
        "strict_passes": null,
        "checks_earned": 24,
        "checks_possible": 31,
        "check_percent": 77.419,
        "wall_s": 139.9,
        "median_decode_tps": 46.532,
        "median_prefill_tps": 22.144,
        "retries": null,
        "truncated": 0,
        "near_quotes": 5,
        "fabricated_quotes": 2,
        "reasoning_fallbacks": 0,
        "failed": 0,
        "scored_cases": 5,
        "retry_count_recorded_cases": 0,
        "request_count_recorded_cases": 0,
        "finish_reason_recorded_cases": 5,
        "cached_prompt_tokens_recorded_cases": 0,
        "elapsed_s_recorded_cases": 5,
        "http_requests": null,
        "metrics": {
          "decode_tps": {
            "n": 5,
            "total": 233.18239174466606,
            "min": 44.598988137835406,
            "median": 46.53209754296622,
            "mean": 46.63647834893322,
            "p95": 48.24905494206578,
            "max": 48.24905494206578
          },
          "prefill_tps": {
            "n": 5,
            "total": 110.11287354926253,
            "min": 21.27048613696066,
            "median": 22.14361239820858,
            "mean": 22.022574709852506,
            "p95": 22.450468653533143,
            "max": 22.450468653533143
          },
          "elapsed_s": {
            "n": 5,
            "total": 139.9,
            "min": 23.9,
            "median": 25.8,
            "mean": 27.98,
            "p95": 35.9,
            "max": 35.9
          },
          "ttft_s": null,
          "prompt_tokens": {
            "n": 5,
            "total": 14442,
            "min": 1834,
            "median": 2850,
            "mean": 2888.4,
            "p95": 3809,
            "max": 3809
          },
          "completion_tokens": {
            "n": 5,
            "total": 6379,
            "min": 1071,
            "median": 1198,
            "mean": 1275.8,
            "p95": 1641,
            "max": 1641
          },
          "cached_prompt_tokens": null
        }
      },
      "groups": {
        "distill": {
          "cases": 5,
          "ok": 5,
          "structure": 4,
          "strict_passes": null,
          "checks_earned": 24,
          "checks_possible": 31,
          "check_percent": 77.419,
          "wall_s": 139.9,
          "median_decode_tps": 46.532,
          "median_prefill_tps": 22.144,
          "retries": null,
          "truncated": 0,
          "near_quotes": 5,
          "fabricated_quotes": 2,
          "reasoning_fallbacks": 0,
          "failed": 0,
          "scored_cases": 5,
          "retry_count_recorded_cases": 0,
          "request_count_recorded_cases": 0,
          "finish_reason_recorded_cases": 5,
          "cached_prompt_tokens_recorded_cases": 0,
          "elapsed_s_recorded_cases": 5,
          "http_requests": null,
          "metrics": {
            "decode_tps": {
              "n": 5,
              "total": 233.18239174466606,
              "min": 44.598988137835406,
              "median": 46.53209754296622,
              "mean": 46.63647834893322,
              "p95": 48.24905494206578,
              "max": 48.24905494206578
            },
            "prefill_tps": {
              "n": 5,
              "total": 110.11287354926253,
              "min": 21.27048613696066,
              "median": 22.14361239820858,
              "mean": 22.022574709852506,
              "p95": 22.450468653533143,
              "max": 22.450468653533143
            },
            "elapsed_s": {
              "n": 5,
              "total": 139.9,
              "min": 23.9,
              "median": 25.8,
              "mean": 27.98,
              "p95": 35.9,
              "max": 35.9
            },
            "ttft_s": null,
            "prompt_tokens": {
              "n": 5,
              "total": 14442,
              "min": 1834,
              "median": 2850,
              "mean": 2888.4,
              "p95": 3809,
              "max": 3809
            },
            "completion_tokens": {
              "n": 5,
              "total": 6379,
              "min": 1071,
              "median": 1198,
              "mean": 1275.8,
              "p95": 1641,
              "max": 1641
            },
            "cached_prompt_tokens": null
          }
        }
      }
    },
    {
      "id": "run-93",
      "run": "q4-r36/distill",
      "model": "Qwen3.8-27B · Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / distill",
      "status": "diagnostic",
      "cases": 2,
      "success": null,
      "quality": null,
      "structure": 2,
      "prefill": 62.25,
      "decode": 13.45,
      "latency": 57.15,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 2,
            "total": 114.3,
            "min": 46.8,
            "median": 57.15,
            "mean": 57.15,
            "max": 67.5
          },
          "prompt_n": {
            "n": 2,
            "total": 1337,
            "min": 554,
            "median": 668.5,
            "mean": 668.5,
            "max": 783
          },
          "pp": {
            "n": 2,
            "total": 124.5,
            "min": 51,
            "median": 62.25,
            "mean": 62.25,
            "max": 73.5
          },
          "tg": {
            "n": 2,
            "total": 26.9,
            "min": 13.1,
            "median": 13.45,
            "mean": 13.45,
            "max": 13.8
          },
          "gen_n": {
            "n": 2,
            "total": 1224,
            "min": 510,
            "median": 612,
            "mean": 612,
            "max": 714
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-94",
      "run": "q4-r36/judge",
      "model": "Qwen3.8-27B · Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / judge",
      "status": "diagnostic",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 129.9,
      "decode": 13.9,
      "latency": 15.6,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 3,
            "total": 53,
            "min": 14.4,
            "median": 15.6,
            "mean": 17.667,
            "max": 23
          },
          "prompt_n": {
            "n": 3,
            "total": 3770,
            "min": 1102,
            "median": 1331,
            "mean": 1256.667,
            "max": 1337
          },
          "pp": {
            "n": 3,
            "total": 347.2,
            "min": 85.2,
            "median": 129.9,
            "mean": 115.733,
            "max": 132.1
          },
          "tg": {
            "n": 3,
            "total": 42.1,
            "min": 13.5,
            "median": 13.9,
            "mean": 14.033,
            "max": 14.7
          },
          "gen_n": {
            "n": 3,
            "total": 268,
            "min": 59,
            "median": 76,
            "mean": 89.333,
            "max": 133
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-95",
      "run": "q4-r36/memory",
      "model": "Qwen3.8-27B · Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / memory",
      "status": "diagnostic",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 82.9,
      "decode": 14.3,
      "latency": 30.8,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 3,
            "total": 92.3,
            "min": 25.9,
            "median": 30.8,
            "mean": 30.767,
            "max": 35.6
          },
          "prompt_n": {
            "n": 3,
            "total": 1685,
            "min": 464,
            "median": 528,
            "mean": 561.667,
            "max": 693
          },
          "pp": {
            "n": 3,
            "total": 284.4,
            "min": 78.7,
            "median": 82.9,
            "mean": 94.8,
            "max": 122.8
          },
          "tg": {
            "n": 3,
            "total": 42.8,
            "min": 14.2,
            "median": 14.3,
            "mean": 14.267,
            "max": 14.3
          },
          "gen_n": {
            "n": 3,
            "total": 1045,
            "min": 315,
            "median": 344,
            "mean": 348.333,
            "max": 386
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-96",
      "run": "q4-r36/reflect",
      "model": "Qwen3.8-27B · Q4_K_XL",
      "model_family": "27B",
      "quant": "Q4_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / reflect",
      "status": "diagnostic",
      "cases": 4,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 47.4,
      "decode": 15.75,
      "latency": 349.2,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 4,
            "total": 1403.2,
            "min": 325.3,
            "median": 349.2,
            "mean": 350.8,
            "max": 379.5
          },
          "prompt_n": {
            "n": 4,
            "total": 17830,
            "min": 3997,
            "median": 4247,
            "mean": 4457.5,
            "max": 5339
          },
          "pp": {
            "n": 4,
            "total": 193.1,
            "min": 35.5,
            "median": 47.4,
            "mean": 48.275,
            "max": 62.8
          },
          "tg": {
            "n": 4,
            "total": 62,
            "min": 14.6,
            "median": 15.75,
            "mean": 15.5,
            "max": 15.9
          },
          "gen_n": {
            "n": 4,
            "total": 15859,
            "min": 3426,
            "median": 4038,
            "mean": 3964.75,
            "max": 4357
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-97",
      "run": "q2-r24/distill",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / distill",
      "status": "diagnostic",
      "cases": 2,
      "success": null,
      "quality": null,
      "structure": 2,
      "prefill": 57.6,
      "decode": 16.05,
      "latency": 48.55,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 2,
            "total": 97.1,
            "min": 35.6,
            "median": 48.55,
            "mean": 48.55,
            "max": 61.5
          },
          "prompt_n": {
            "n": 2,
            "total": 1337,
            "min": 554,
            "median": 668.5,
            "mean": 668.5,
            "max": 783
          },
          "pp": {
            "n": 2,
            "total": 115.2,
            "min": 44.2,
            "median": 57.6,
            "mean": 57.6,
            "max": 71
          },
          "tg": {
            "n": 2,
            "total": 32.1,
            "min": 15.5,
            "median": 16.05,
            "mean": 16.05,
            "max": 16.6
          },
          "gen_n": {
            "n": 2,
            "total": 1147,
            "min": 429,
            "median": 573.5,
            "mean": 573.5,
            "max": 718
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-98",
      "run": "q2-r24/judge",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / judge",
      "status": "diagnostic",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 173.5,
      "decode": 17.2,
      "latency": 12.5,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 3,
            "total": 42.5,
            "min": 10.4,
            "median": 12.5,
            "mean": 14.167,
            "max": 19.6
          },
          "prompt_n": {
            "n": 3,
            "total": 3770,
            "min": 1102,
            "median": 1331,
            "mean": 1256.667,
            "max": 1337
          },
          "pp": {
            "n": 3,
            "total": 433.7,
            "min": 81.5,
            "median": 173.5,
            "mean": 144.567,
            "max": 178.7
          },
          "tg": {
            "n": 3,
            "total": 51.7,
            "min": 15.8,
            "median": 17.2,
            "mean": 17.233,
            "max": 18.7
          },
          "gen_n": {
            "n": 3,
            "total": 227,
            "min": 48,
            "median": 87,
            "mean": 75.667,
            "max": 92
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-99",
      "run": "q2-r24/memory",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / memory",
      "status": "diagnostic",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 79.5,
      "decode": 16.7,
      "latency": 24.9,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 3,
            "total": 70.6,
            "min": 17.1,
            "median": 24.9,
            "mean": 23.533,
            "max": 28.6
          },
          "prompt_n": {
            "n": 3,
            "total": 1685,
            "min": 464,
            "median": 528,
            "mean": 561.667,
            "max": 693
          },
          "pp": {
            "n": 3,
            "total": 323.1,
            "min": 74.9,
            "median": 79.5,
            "mean": 107.7,
            "max": 168.7
          },
          "tg": {
            "n": 3,
            "total": 50.7,
            "min": 16.7,
            "median": 16.7,
            "mean": 16.9,
            "max": 17.3
          },
          "gen_n": {
            "n": 3,
            "total": 874,
            "min": 236,
            "median": 296,
            "mean": 291.333,
            "max": 342
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-100",
      "run": "q2-r24/reflect",
      "model": "Qwen3.8-27B · Q2_K_XL",
      "model_family": "27B",
      "quant": "Q2_K_XL",
      "runtime": "llama.cpp CUDA · early quality",
      "suite": "early quality / reflect",
      "status": "diagnostic",
      "cases": 4,
      "success": null,
      "quality": null,
      "structure": 3,
      "prefill": 132.8,
      "decode": 22.05,
      "latency": 231.85,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Format passes and separate judge verdicts; not public123/frozen24 scoring. HTTP success is not recorded.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "wall": {
            "n": 4,
            "total": 944.8,
            "min": 193.7,
            "median": 231.85,
            "mean": 236.2,
            "max": 287.4
          },
          "prompt_n": {
            "n": 4,
            "total": 17830,
            "min": 3997,
            "median": 4247,
            "mean": 4457.5,
            "max": 5339
          },
          "pp": {
            "n": 4,
            "total": 518.2,
            "min": 61.4,
            "median": 132.8,
            "mean": 129.55,
            "max": 191.2
          },
          "tg": {
            "n": 4,
            "total": 87.9,
            "min": 21.1,
            "median": 22.05,
            "mean": 21.975,
            "max": 22.7
          },
          "gen_n": {
            "n": 4,
            "total": 16665,
            "min": 3643,
            "median": 4260,
            "mean": 4166.25,
            "max": 4502
          }
        }
      },
      "groups": {}
    },
    {
      "id": "run-101",
      "run": "32K/one-slot kernel tuning, best of three warm, 1201 prompt/256 output · MTP",
      "model": "27B Q2",
      "model_family": "27B",
      "quant": "Q2",
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": 706,
      "decode": 40,
      "latency": null,
      "retries": null,
      "truncated": null,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-102",
      "run": "32K/one-slot kernel tuning, best of three warm, 1201 prompt/256 output · no speculation",
      "model": "27B Q2",
      "model_family": "27B",
      "quant": "Q2",
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": 774,
      "decode": 25.1,
      "latency": null,
      "retries": null,
      "truncated": null,
      "date": "historical · date not in filename",
      "gpu_count": null,
      "gpu_used": "Unknown / unrecorded",
      "placement_basis": "GPU count is not recorded in recovered run-specific evidence.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-103",
      "run": "CPU-only llama-bench, September 6; not rerun",
      "model": "Flash Next (quant unrecorded)",
      "model_family": "Flash Next",
      "quant": null,
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": 63,
      "decode": 4.5,
      "latency": null,
      "retries": null,
      "truncated": null,
      "date": "historical · date not in filename",
      "gpu_count": 0,
      "gpu_used": "CPU only · 0 GPUs",
      "placement_basis": "Recorded CPU-only benchmark.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-104",
      "run": "direct context capacity probe",
      "model": "27B Q2",
      "model_family": "27B",
      "quant": "Q2",
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": 400.48,
      "decode": 14.11,
      "latency": 301.278,
      "retries": 0,
      "truncated": 0,
      "date": "20260917T154912Z",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "retries": 0
      },
      "groups": {}
    },
    {
      "id": "run-105",
      "run": "gateway context capacity probe",
      "model": "27B Q2",
      "model_family": "27B",
      "quant": "Q2",
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": null,
      "decode": null,
      "latency": 307.897,
      "retries": null,
      "truncated": null,
      "date": "20260917T154912Z",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-106",
      "run": "direct context capacity probe",
      "model": "9B Q4",
      "model_family": "9B",
      "quant": "Q4",
      "runtime": "reported-only; see source",
      "suite": "reported-only probe",
      "status": "reported-only",
      "cases": null,
      "success": null,
      "quality": null,
      "structure": null,
      "prefill": 1326.48,
      "decode": 26.86,
      "latency": 90.822,
      "retries": 0,
      "truncated": 0,
      "date": "20260917T165935Z",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Archived validation.md specifies one GPU.",
      "notes": "Historical reported result; independent raw artifact and sample count are unavailable.",
      "hardware": "Xeon · dual RTX 4060 Ti",
      "summary": {
        "retries": 0
      },
      "groups": {}
    },
    {
      "id": "run-107",
      "run": "20260930T232800Z-b70-27b-q4/streaming-ttft",
      "model": "Qwen3.8-27B-UD-Q4_K_XL",
      "model_family": "27B",
      "quant": "UD-Q4_K_XL",
      "runtime": "llama.cpp SYCL",
      "suite": "streaming TTFT probe",
      "status": "probe",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": null,
      "ttft": 1.4428602039988618,
      "latency": 6.060961260998738,
      "cache": 0,
      "context": 131072,
      "slots": 2,
      "kv": "q8_0",
      "date": "20260930T232800Z",
      "gpu_count": 1,
      "gpu_used": "1 × Arc Pro B70 32 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "Separate short synthetic uncached probes; not part of public123 quality scoring.",
      "hardware": "Core i5 · Arc Pro B70",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-108",
      "run": "TTS · cold first request",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.22680172599939397,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "One fixed phrase, 2.197 seconds of audio. Transcription correctness is not broad WER; no general quality eval.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "n": {},
          "total": {},
          "min": {},
          "median": {},
          "mean": {},
          "p95": {},
          "max": {}
        }
      },
      "groups": {}
    },
    {
      "id": "run-109",
      "run": "STT · cold first request",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": 100,
      "quality": null,
      "structure": null,
      "latency": 0.2160395700011577,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "One fixed phrase, 2.197 seconds of audio. Transcription correctness is not broad WER; no general quality eval.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "n": {},
          "total": {},
          "min": {},
          "median": {},
          "mean": {},
          "p95": {},
          "max": {}
        }
      },
      "groups": {}
    },
    {
      "id": "run-110",
      "run": "TTS · warm direct HTTP",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.07336530400061747,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "One fixed phrase, 2.197 seconds of audio. Transcription correctness is not broad WER; no general quality eval.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "n": {},
          "total": {},
          "min": {},
          "median": {},
          "mean": {},
          "p95": {},
          "max": {}
        }
      },
      "groups": {}
    },
    {
      "id": "run-111",
      "run": "STT · warm direct HTTP",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": 100,
      "quality": null,
      "structure": null,
      "latency": 0.1229186229975312,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "One fixed phrase, 2.197 seconds of audio. Transcription correctness is not broad WER; no general quality eval.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {
        "metrics": {
          "n": {},
          "total": {},
          "min": {},
          "median": {},
          "mean": {},
          "p95": {},
          "max": {}
        }
      },
      "groups": {}
    },
    {
      "id": "run-112",
      "run": "TTS · HTTPS gateway",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.11800988204777241,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Three calls including the first slower call; fixed 2.197-second phrase. Not an independent accuracy benchmark.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-113",
      "run": "STT · HTTPS gateway",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 3,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.16599131911061704,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Three calls including the first slower call; fixed 2.197-second phrase. Not an independent accuracy benchmark.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-114",
      "run": "TTS · concurrent speech + embedding",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.12437907187268138,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Single concurrent smoke; not sustained throughput or contention benchmark.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-115",
      "run": "STT · concurrent speech + embedding",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.1879048920236528,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Single concurrent smoke; not sustained throughput or contention benchmark.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-116",
      "run": "Embedding · concurrent speech + embedding",
      "model": "Qwen3-Embedding-0.6B Q8_0",
      "model_family": "Qwen3-Embedding-0.6B Q8_0",
      "quant": "Q8_0",
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.057482759933918715,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Single concurrent smoke; not sustained throughput or contention benchmark.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-117",
      "run": "ComfyUI · SDXL image generation",
      "model": "SDXL",
      "model_family": "SDXL",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": 100,
      "quality": null,
      "structure": null,
      "latency": 27.75969087300473,
      "date": "historical · date not in filename",
      "gpu_count": 1,
      "gpu_used": "1 × Arc A770 16 GB",
      "placement_basis": "Recorded single B70 / A770 / RTX 5090 benchmark accelerator.",
      "notes": "SDXL 512x512, 8 Euler steps; smoke test, not a production speed benchmark",
      "hardware": "Core i5 · Arc A770",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-118",
      "run": "Intern-Decision-2B · first decision",
      "model": "Intern-Decision-2B · BF16",
      "model_family": "Intern-Decision-2B",
      "quant": null,
      "runtime": "Hugging Face · PyTorch CUDA · BF16",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.2914513150026323,
      "context": 8192,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "276 input tokens, three decision fields; first-call warmup. Non-autoregressive decision latency; decode/prefill t/s are not applicable.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-119",
      "run": "Intern-Decision-2B · warm short decisions",
      "model": "Intern-Decision-2B · BF16",
      "model_family": "Intern-Decision-2B",
      "quant": null,
      "runtime": "Hugging Face · PyTorch CUDA · BF16",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 12,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.06585767400247278,
      "context": 8192,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "276 input tokens, three fields, same toy decision repeated. Not broad accuracy. Non-autoregressive decision latency; decode/prefill t/s are not applicable.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-120",
      "run": "Intern-Decision-2B · concurrent short decisions",
      "model": "Intern-Decision-2B · BF16",
      "model_family": "Intern-Decision-2B",
      "quant": null,
      "runtime": "Hugging Face · PyTorch CUDA · BF16",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.13414304099569563,
      "context": 8192,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Five rounds alongside TTS/STT and eight embedding inputs. Non-autoregressive decision latency; decode/prefill t/s are not applicable.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-121",
      "run": "Intern-Decision-2B · long-input concurrent decision",
      "model": "Intern-Decision-2B · BF16",
      "model_family": "Intern-Decision-2B",
      "quant": null,
      "runtime": "Hugging Face · PyTorch CUDA · BF16",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 1.6991615079969051,
      "context": 8192,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "7,989 input tokens, one field, alongside speech and embeddings. Single sample. Non-autoregressive decision latency; decode/prefill t/s are not applicable.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-122",
      "run": "TTS · warm speech after decision load",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.07349519499985036,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-123",
      "run": "STT · warm speech after decision load",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.12280819300212897,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-124",
      "run": "TTS · concurrent with short decision",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.1806639450005605,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-125",
      "run": "STT · concurrent with short decision",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.2016164749948075,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-126",
      "run": "Embedding · concurrent with short decision",
      "model": "Qwen3-Embedding-0.6B Q8_0",
      "model_family": "Qwen3-Embedding-0.6B Q8_0",
      "quant": "Q8_0",
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 5,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.16578958700119983,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Eight inputs per batch alongside decision and speech; smoke, not sustained throughput.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-127",
      "run": "TTS · concurrent with long decision",
      "model": "Kokoro-82M ONNX",
      "model_family": "Kokoro-82M ONNX",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.21381652900163317,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-128",
      "run": "STT · concurrent with long decision",
      "model": "faster-whisper-small.en",
      "model_family": "faster-whisper-small.en",
      "quant": null,
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.5790697890042793,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Fixed short audio, decision GPU coexistence smoke; not broad speech quality or sustained concurrency.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    },
    {
      "id": "run-129",
      "run": "Embedding · concurrent with long decision",
      "model": "Qwen3-Embedding-0.6B Q8_0",
      "model_family": "Qwen3-Embedding-0.6B Q8_0",
      "quant": "Q8_0",
      "runtime": "recorded serving smoke",
      "suite": "service smoke · not LLM quality eval",
      "status": "smoke",
      "cases": 1,
      "success": null,
      "quality": null,
      "structure": null,
      "latency": 0.27272426600393374,
      "date": "2026-10-01T02:05:13.187948+00:00",
      "gpu_count": 1,
      "gpu_used": "1 × RTX 4060 Ti 16 GB",
      "placement_basis": "Recorded CUDA service deployment on GPU1; speech, embeddings and decision share that card.",
      "notes": "Eight inputs per batch alongside decision and speech; smoke, not sustained throughput.",
      "hardware": "Ryzen · dual RTX 4060 Ti",
      "summary": {},
      "groups": {}
    }
  ]
}
