{
  "schemaVersion": 1,
  "modelPath": "/Users/brooklyn/workspace/github/mlx-node/.cache/models/ornith-1.0-35b-unsloth-mxfp4-mlx",
  "startedAt": "2026-07-20T16:43:42.832Z",
  "completedAt": "2026-07-20T16:54:10.310Z",
  "environment": {
    "platform": "darwin",
    "arch": "arm64",
    "osRelease": "25.5.0",
    "osVersion": "Darwin Kernel Version 25.5.0: Tue Jun  9 22:28:34 PDT 2026; root:xnu-12377.121.10~1/RELEASE_ARM64_T6050",
    "cpu": "Apple M5 Max",
    "logicalCpus": 18,
    "totalMemoryBytes": 137438953472,
    "nodeVersion": "v24.13.1",
    "packages": {
      "lm": {
        "name": "@mlx-node/lm",
        "version": "0.0.8"
      },
      "core": {
        "name": "@mlx-node/core",
        "version": "0.0.8"
      },
      "native": {
        "name": "@mlx-node/core-darwin-arm64",
        "version": "0.0.8"
      }
    }
  },
  "configuration": {
    "runs": 3,
    "maxNewTokens": 512,
    "cooldownSeconds": 60,
    "warmupRuns": 0,
    "temperature": 0,
    "reasoningEffort": "none",
    "reportPerformance": true,
    "minimumCompletionRatio": 0.95,
    "systemPrompt": "You are a technical writer in a deterministic inference benchmark. Produce a continuous, detailed response and do not conclude early.",
    "prompt": "Write a comprehensive technical handbook chapter about building a production compiler. Cover lexical analysis, parsing, semantic analysis, type checking, intermediate representations, optimization, code generation, linking, testing, debugging, and deployment. Use detailed explanations and concrete examples. Do not summarize or conclude early; keep expanding the chapter until the generation limit is reached.",
    "isolation": "fresh child process and model load per sample"
  },
  "warmups": [],
  "runs": [
    {
      "pid": 4948,
      "loadMs": 295858.866667,
      "generationWallMs": 8295.485790999955,
      "totalWallMs": 304154.46933299996,
      "ttftMs": 473.89941699999997,
      "prefillTokensPerSecond": 223.6761561578372,
      "decodeTokensPerSecond": 67.44062557118689,
      "promptTokens": 106,
      "generatedTokens": 512,
      "reasoningTokens": 0,
      "cachedTokens": 0,
      "finishReason": "length",
      "minimumGeneratedTokens": 487,
      "phase": "measured",
      "run": 1
    },
    {
      "pid": 14146,
      "loadMs": 77110.606125,
      "generationWallMs": 6654.497124999994,
      "totalWallMs": 83765.210292,
      "ttftMs": 361.292125,
      "prefillTokensPerSecond": 293.39139346034733,
      "decodeTokensPerSecond": 82.95027113210845,
      "promptTokens": 106,
      "generatedTokens": 512,
      "reasoningTokens": 0,
      "cachedTokens": 0,
      "finishReason": "length",
      "minimumGeneratedTokens": 487,
      "phase": "measured",
      "run": 2
    },
    {
      "pid": 17338,
      "loadMs": 111887.36429200001,
      "generationWallMs": 6649.128375,
      "totalWallMs": 118536.607042,
      "ttftMs": 364.8845,
      "prefillTokensPerSecond": 290.5028851595505,
      "decodeTokensPerSecond": 83.18705959265836,
      "promptTokens": 106,
      "generatedTokens": 512,
      "reasoningTokens": 0,
      "cachedTokens": 0,
      "finishReason": "length",
      "minimumGeneratedTokens": 487,
      "phase": "measured",
      "run": 3
    }
  ],
  "medians": {
    "loadMs": 111887.36429200001,
    "generationWallMs": 6654.497124999994,
    "totalWallMs": 118536.607042,
    "ttftMs": 364.8845,
    "prefillTokensPerSecond": 290.5028851595505,
    "decodeTokensPerSecond": 82.95027113210845,
    "promptTokens": 106,
    "generatedTokens": 512
  }
}