{
  "task": "鵜鶘騎腳踏車 SVG 動畫",
  "context": 262144,
  "reasoning": "xhigh",
  "freshSessionsPerModel": 1,
  "supervisorRepairs": 0,
  "results": [
    {
      "name": "Bonsai 2 27B",
      "configuration": "PTQ1_0 · 本機 RTX 4080 SUPER",
      "path": "bonsai-ptq/",
      "sha256": "f55fd73c6b39c3cb6a6449f3273844c93fa9bbd741c55c191776d52a0718a8b0",
      "elapsedSeconds": 3264.40333,
      "exitCode": 0,
      "checksPassed": 18,
      "checksTotal": 30,
      "failedChecks": [
        "1440_correct_title",
        "1440_animation_running",
        "1440_resume_animates",
        "1440_speed_doubles",
        "390_correct_title",
        "390_no_overflow",
        "390_rider_visible",
        "390_animation_running",
        "390_resume_animates",
        "390_speed_doubles",
        "reduced_motion_manual_play",
        "no_browser_errors"
      ],
      "browserErrors": [
        "Cannot read properties of null (reading 'setAttribute')",
        "Failed to load resource: the server responded with a status of 404 (File not found)",
        "Page.evaluate: TypeError: Cannot read properties of null (reading 'setAttribute')",
        "Cannot read properties of null (reading 'setAttribute')",
        "Cannot read properties of null (reading 'setAttribute')"
      ],
      "artifactCreated": true,
      "finishReason": "stop",
      "outputTokens": 127667,
      "reasoningTokens": 0,
      "opencodeEstimatedCostUSD": 0,
      "acceptancePassed": false
    },
    {
      "name": "Qwen3.8 27B",
      "configuration": "BF16 · OpenRouter / DeepInfra",
      "path": null,
      "sha256": null,
      "elapsedSeconds": 2043.9880448,
      "exitCode": 0,
      "checksPassed": 0,
      "checksTotal": 0,
      "failedChecks": [],
      "browserErrors": [],
      "artifactCreated": false,
      "finishReason": "length",
      "outputTokens": 22786,
      "reasoningTokens": 42750,
      "opencodeEstimatedCostUSD": 0.169467376,
      "acceptancePassed": false
    }
  ],
  "note": "單次觀察，不代表模型整體能力或單獨量化造成的差異。瀏覽器檢查是功能驗收，不是美術評分。"
}