{
  "capturedAt": "2026-07-13T15:14:33.920Z",
  "environment": {
    "platform": "darwin",
    "arch": "arm64",
    "release": "25.5.0",
    "cpuModel": "Apple M4",
    "logicalCpuCount": 10,
    "totalMemoryBytes": 17179869184
  },
  "model": {
    "name": "qwen3:8b",
    "digest": "500a1f067a9f782620b40bee6f7b0c89e17ae61f686b92c24933e4ca4b2b8b41",
    "size": 5225388164,
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "qwen3",
      "families": [
        "qwen3"
      ],
      "parameter_size": "8.2B",
      "quantization_level": "Q4_K_M",
      "context_length": 40960,
      "embedding_length": 4096
    }
  },
  "settings": {
    "rounds": 2,
    "temperature": 0,
    "seed": 42,
    "numCtx": 4096,
    "numPredict": 96,
    "think": false
  },
  "summary": {
    "requests": 14,
    "completed": 14,
    "taskAccuracy": 0.8571428571428571,
    "ttftMs": {
      "p50": 178.995,
      "p95": 184.003
    },
    "totalMs": {
      "p50": 636.774,
      "p95": 1869.235
    },
    "outputTokensPerSecond": {
      "p50": 14.98,
      "p95": 15.828
    }
  },
  "runs": [
    {
      "round": 1,
      "taskId": "incident-classification-ambiguous",
      "correct": false,
      "response": "{\"severity\":\"P1 或 P2\"}",
      "ttftMs": 178.564,
      "totalMs": 871.706,
      "promptEvalCount": 45,
      "outputTokenCount": 10,
      "outputTokensPerSecond": 14.405,
      "loadMs": 96.479
    },
    {
      "round": 1,
      "taskId": "incident-classification-constrained",
      "correct": true,
      "response": "{\"severity\":\"P1\"}",
      "ttftMs": 175.833,
      "totalMs": 636.774,
      "promptEvalCount": 60,
      "outputTokenCount": 7,
      "outputTokensPerSecond": 15.196,
      "loadMs": 94.043
    },
    {
      "round": 1,
      "taskId": "invoice-extraction",
      "correct": true,
      "response": "{\"tax_id\":\"91310000MA1K123456\"}",
      "ttftMs": 184.003,
      "totalMs": 1869.235,
      "promptEvalCount": 72,
      "outputTokenCount": 23,
      "outputTokensPerSecond": 13.649,
      "loadMs": 95.279
    },
    {
      "round": 1,
      "taskId": "code-review",
      "correct": true,
      "response": "{\"bug\":\"assignment\"}",
      "ttftMs": 180.409,
      "totalMs": 562.894,
      "promptEvalCount": 45,
      "outputTokenCount": 6,
      "outputTokensPerSecond": 15.685,
      "loadMs": 93.725
    },
    {
      "round": 1,
      "taskId": "policy-lookup",
      "correct": true,
      "response": "{\"minutes\":10}",
      "ttftMs": 181.96,
      "totalMs": 636.163,
      "promptEvalCount": 74,
      "outputTokenCount": 7,
      "outputTokensPerSecond": 15.389,
      "loadMs": 94.181
    },
    {
      "round": 1,
      "taskId": "tool-approval",
      "correct": true,
      "response": "{\"decision\":\"block\"}",
      "ttftMs": 181.153,
      "totalMs": 563.693,
      "promptEvalCount": 51,
      "outputTokenCount": 6,
      "outputTokensPerSecond": 15.674,
      "loadMs": 95.571
    },
    {
      "round": 1,
      "taskId": "arithmetic-check",
      "correct": true,
      "response": "{\"answer\":391}",
      "ttftMs": 183.498,
      "totalMs": 718.584,
      "promptEvalCount": 34,
      "outputTokenCount": 8,
      "outputTokensPerSecond": 14.98,
      "loadMs": 100.747
    },
    {
      "round": 2,
      "taskId": "incident-classification-ambiguous",
      "correct": false,
      "response": "{\"severity\":\"P1 或 P2\"}",
      "ttftMs": 178.87,
      "totalMs": 899.781,
      "promptEvalCount": 45,
      "outputTokenCount": 10,
      "outputTokensPerSecond": 13.871,
      "loadMs": 93.871
    },
    {
      "round": 2,
      "taskId": "incident-classification-constrained",
      "correct": true,
      "response": "{\"severity\":\"P1\"}",
      "ttftMs": 178.995,
      "totalMs": 651.849,
      "promptEvalCount": 60,
      "outputTokenCount": 7,
      "outputTokensPerSecond": 14.804,
      "loadMs": 94.854
    },
    {
      "round": 2,
      "taskId": "invoice-extraction",
      "correct": true,
      "response": "{\"tax_id\":\"91310000MA1K123456\"}",
      "ttftMs": 180.573,
      "totalMs": 1865.252,
      "promptEvalCount": 72,
      "outputTokenCount": 23,
      "outputTokensPerSecond": 13.656,
      "loadMs": 96.502
    },
    {
      "round": 2,
      "taskId": "code-review",
      "correct": true,
      "response": "{\"bug\":\"assignment\"}",
      "ttftMs": 178.85,
      "totalMs": 558.655,
      "promptEvalCount": 45,
      "outputTokenCount": 6,
      "outputTokensPerSecond": 15.828,
      "loadMs": 93.927
    },
    {
      "round": 2,
      "taskId": "policy-lookup",
      "correct": true,
      "response": "{\"minutes\":10}",
      "ttftMs": 174.863,
      "totalMs": 633.821,
      "promptEvalCount": 74,
      "outputTokenCount": 7,
      "outputTokensPerSecond": 15.271,
      "loadMs": 96.099
    },
    {
      "round": 2,
      "taskId": "tool-approval",
      "correct": true,
      "response": "{\"decision\":\"block\"}",
      "ttftMs": 173.059,
      "totalMs": 557.516,
      "promptEvalCount": 51,
      "outputTokenCount": 6,
      "outputTokensPerSecond": 15.617,
      "loadMs": 95.38
    },
    {
      "round": 2,
      "taskId": "arithmetic-check",
      "correct": true,
      "response": "{\"answer\":391}",
      "ttftMs": 179.548,
      "totalMs": 716.387,
      "promptEvalCount": 34,
      "outputTokenCount": 8,
      "outputTokensPerSecond": 14.902,
      "loadMs": 95.632
    }
  ]
}
