{
 "version": "0.1.0",
 "date": "2026-10-17",
 "machine": "AMD Ryzen AI 5 PRO 340 w/ Radeon 840M, Linux 6.17.0-1033-oem",
 "variant": "default (slim prompt)",
 "scenarios": [
  {
   "id": "fix_qos",
   "title": "Fix two silent QoS problems",
   "checks": 2
  },
  {
   "id": "cloud_height",
   "title": "Add a 3D cloud colored by height",
   "checks": 4
  },
  {
   "id": "tf_robot",
   "title": "TF error + robot model + QoS + 3D cloud (the video request)",
   "checks": 6
  },
  {
   "id": "follow_up",
   "title": "Follow-up: \"make the points bigger\"",
   "checks": 1
  },
  {
   "id": "nav2_view",
   "title": "Show what Nav2 is planning",
   "checks": 5
  },
  {
   "id": "light_weight",
   "title": "Make a heavy cloud light-weight (and be honest about it)",
   "checks": 2
  },
  {
   "id": "camera",
   "title": "Pick the real camera topic",
   "checks": 1
  },
  {
   "id": "hide_rename",
   "title": "Hide and rename",
   "checks": 2
  }
 ],
 "results": [
  {
   "provider": "anthropic",
   "model": "claude-opus-5-5",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 5.38,
     "usage": {
      "prompt_tokens": 2719,
      "output_tokens": 251,
      "model_seconds": 5.38,
      "prompt_chars": 3340
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 3.55,
     "usage": {
      "prompt_tokens": 1799,
      "output_tokens": 206,
      "model_seconds": 3.54,
      "prompt_chars": 1273
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 4.92,
     "usage": {
      "prompt_tokens": 3386,
      "output_tokens": 473,
      "model_seconds": 4.92,
      "prompt_chars": 4886
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      }
     ],
     "passed": 6,
     "total": 6,
     "rejected": 0
    },
    {
     "scenario": "follow_up",
     "seconds": 3.37,
     "usage": {
      "prompt_tokens": 2145,
      "output_tokens": 129,
      "model_seconds": 3.37,
      "prompt_chars": 2015
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 7.1,
     "usage": {
      "prompt_tokens": 2498,
      "output_tokens": 704,
      "model_seconds": 7.1,
      "prompt_chars": 2944
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": true
      },
      {
       "label": "local costmap",
       "ok": true
      },
      {
       "label": "costmap color scheme",
       "ok": true
      },
      {
       "label": "footprint polygon",
       "ok": true
      }
     ],
     "passed": 5,
     "total": 5,
     "rejected": 0
    },
    {
     "scenario": "light_weight",
     "seconds": 5.71,
     "usage": {
      "prompt_tokens": 2051,
      "output_tokens": 364,
      "model_seconds": 5.7,
      "prompt_chars": 1798
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "camera",
     "seconds": 2.7,
     "usage": {
      "prompt_tokens": 1696,
      "output_tokens": 133,
      "model_seconds": 2.7,
      "prompt_chars": 1069
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 4.62,
     "usage": {
      "prompt_tokens": 3283,
      "output_tokens": 247,
      "model_seconds": 4.62,
      "prompt_chars": 4533
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": true
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 8,
   "scenarios": 8,
   "checks_passed": 23,
   "checks": 23,
   "errors": 0,
   "rejected": 0,
   "median_seconds": 4.8,
   "mean_prompt_tokens": 2447,
   "mean_output_tokens": 313,
   "mean_prompt_seconds": null,
   "mean_output_seconds": null
  },
  {
   "provider": "anthropic",
   "model": "claude-sonnet-5-5",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 4.05,
     "usage": {
      "prompt_tokens": 2719,
      "output_tokens": 168,
      "model_seconds": 4.05,
      "prompt_chars": 3340
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 3.76,
     "usage": {
      "prompt_tokens": 1799,
      "output_tokens": 203,
      "model_seconds": 3.76,
      "prompt_chars": 1273
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 3.77,
     "usage": {
      "prompt_tokens": 3386,
      "output_tokens": 513,
      "model_seconds": 3.77,
      "prompt_chars": 4886
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      }
     ],
     "passed": 6,
     "total": 6,
     "rejected": 0
    },
    {
     "scenario": "follow_up",
     "seconds": 1.93,
     "usage": {
      "prompt_tokens": 2145,
      "output_tokens": 127,
      "model_seconds": 1.93,
      "prompt_chars": 2015
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 4.77,
     "usage": {
      "prompt_tokens": 2498,
      "output_tokens": 697,
      "model_seconds": 4.77,
      "prompt_chars": 2944
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": true
      },
      {
       "label": "local costmap",
       "ok": true
      },
      {
       "label": "costmap color scheme",
       "ok": true
      },
      {
       "label": "footprint polygon",
       "ok": true
      }
     ],
     "passed": 5,
     "total": 5,
     "rejected": 0
    },
    {
     "scenario": "light_weight",
     "seconds": 3.06,
     "usage": {
      "prompt_tokens": 2051,
      "output_tokens": 341,
      "model_seconds": 3.06,
      "prompt_chars": 1798
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "camera",
     "seconds": 2.27,
     "usage": {
      "prompt_tokens": 1696,
      "output_tokens": 97,
      "model_seconds": 2.27,
      "prompt_chars": 1069
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 3.59,
     "usage": {
      "prompt_tokens": 3283,
      "output_tokens": 228,
      "model_seconds": 3.59,
      "prompt_chars": 4533
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": true
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 8,
   "scenarios": 8,
   "checks_passed": 23,
   "checks": 23,
   "errors": 0,
   "rejected": 0,
   "median_seconds": 3.7,
   "mean_prompt_tokens": 2447,
   "mean_output_tokens": 297,
   "mean_prompt_seconds": null,
   "mean_output_seconds": null
  },
  {
   "provider": "anthropic",
   "model": "claude-haiku-5-5",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 2.13,
     "usage": {
      "prompt_tokens": 2719,
      "output_tokens": 267,
      "model_seconds": 2.13,
      "prompt_chars": 3340
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 2.17,
     "usage": {
      "prompt_tokens": 1799,
      "output_tokens": 235,
      "model_seconds": 2.17,
      "prompt_chars": 1273
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 2.98,
     "usage": {
      "prompt_tokens": 3386,
      "output_tokens": 559,
      "model_seconds": 2.98,
      "prompt_chars": 4886
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      }
     ],
     "passed": 6,
     "total": 6,
     "rejected": 0
    },
    {
     "scenario": "follow_up",
     "seconds": 2.04,
     "usage": {
      "prompt_tokens": 2145,
      "output_tokens": 168,
      "model_seconds": 2.04,
      "prompt_chars": 2015
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 4.49,
     "usage": {
      "prompt_tokens": 2498,
      "output_tokens": 1041,
      "model_seconds": 4.49,
      "prompt_chars": 2944
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": true
      },
      {
       "label": "local costmap",
       "ok": true
      },
      {
       "label": "costmap color scheme",
       "ok": true
      },
      {
       "label": "footprint polygon",
       "ok": true
      }
     ],
     "passed": 5,
     "total": 5,
     "rejected": 0
    },
    {
     "scenario": "light_weight",
     "seconds": 3.63,
     "usage": {
      "prompt_tokens": 2051,
      "output_tokens": 408,
      "model_seconds": 3.62,
      "prompt_chars": 1798
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "camera",
     "seconds": 1.72,
     "usage": {
      "prompt_tokens": 1696,
      "output_tokens": 133,
      "model_seconds": 1.72,
      "prompt_chars": 1069
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 1.99,
     "usage": {
      "prompt_tokens": 3283,
      "output_tokens": 205,
      "model_seconds": 1.99,
      "prompt_chars": 4533
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": true
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 8,
   "scenarios": 8,
   "checks_passed": 23,
   "checks": 23,
   "errors": 0,
   "rejected": 0,
   "median_seconds": 2.1,
   "mean_prompt_tokens": 2447,
   "mean_output_tokens": 377,
   "mean_prompt_seconds": null,
   "mean_output_seconds": null
  },
  {
   "provider": "ollama",
   "model": "qwen3:4b-instruct",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 24.94,
     "usage": {
      "prompt_tokens": 1357,
      "output_tokens": 202,
      "prompt_seconds": 11.169224,
      "output_seconds": 13.635012,
      "load_seconds": 0.000875855,
      "model_seconds": 24.94,
      "prompt_chars": 3340
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 30.71,
     "usage": {
      "prompt_tokens": 811,
      "output_tokens": 446,
      "prompt_seconds": 3.547189,
      "output_seconds": 27.058761,
      "load_seconds": 0.000863631,
      "model_seconds": 30.71,
      "prompt_chars": 1273
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 70.74,
     "usage": {
      "prompt_tokens": 1782,
      "output_tokens": 755,
      "prompt_seconds": 17.29739,
      "output_seconds": 53.352385,
      "load_seconds": 0.00094345,
      "model_seconds": 70.74,
      "prompt_chars": 4886
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      }
     ],
     "passed": 6,
     "total": 6,
     "rejected": 0
    },
    {
     "scenario": "follow_up",
     "seconds": 14.51,
     "usage": {
      "prompt_tokens": 1010,
      "output_tokens": 135,
      "prompt_seconds": 5.938833,
      "output_seconds": 8.397674,
      "load_seconds": 0.000838946,
      "model_seconds": 14.51,
      "prompt_chars": 2015
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 44.21,
     "usage": {
      "prompt_tokens": 1230,
      "output_tokens": 548,
      "prompt_seconds": 8.822003,
      "output_seconds": 35.312065,
      "load_seconds": 0.000860737,
      "model_seconds": 44.21,
      "prompt_chars": 2944
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": true
      },
      {
       "label": "local costmap",
       "ok": true
      },
      {
       "label": "costmap color scheme",
       "ok": true
      },
      {
       "label": "footprint polygon",
       "ok": true
      }
     ],
     "passed": 5,
     "total": 5,
     "rejected": 0
    },
    {
     "scenario": "light_weight",
     "seconds": 43.84,
     "usage": {
      "prompt_tokens": 954,
      "output_tokens": 696,
      "prompt_seconds": 0.562155,
      "output_seconds": 43.153377,
      "load_seconds": 0.000999225,
      "model_seconds": 43.84,
      "prompt_chars": 1798
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "camera",
     "seconds": 28.02,
     "usage": {
      "prompt_tokens": 734,
      "output_tokens": 426,
      "prompt_seconds": 2.437032,
      "output_seconds": 25.489383,
      "load_seconds": 0.000875224,
      "model_seconds": 28.02,
      "prompt_chars": 1069
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 33.41,
     "usage": {
      "prompt_tokens": 1720,
      "output_tokens": 255,
      "prompt_seconds": 15.913686,
      "output_seconds": 17.413689,
      "load_seconds": 0.001131121,
      "model_seconds": 33.41,
      "prompt_chars": 4533
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": true
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 8,
   "scenarios": 8,
   "checks_passed": 23,
   "checks": 23,
   "errors": 0,
   "rejected": 0,
   "median_seconds": 32.1,
   "mean_prompt_tokens": 1200,
   "mean_output_tokens": 433,
   "mean_prompt_seconds": 8.2,
   "mean_output_seconds": 28.0
  },
  {
   "provider": "ollama",
   "model": "qwen2.5:3b",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 23.96,
     "usage": {
      "prompt_tokens": 1357,
      "output_tokens": 175,
      "prompt_seconds": 12.601271,
      "output_seconds": 8.398102999,
      "load_seconds": 2.84222401,
      "model_seconds": 23.96,
      "prompt_chars": 3340
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 22.38,
     "usage": {
      "prompt_tokens": 811,
      "output_tokens": 434,
      "prompt_seconds": 2.616423,
      "output_seconds": 19.739247,
      "load_seconds": 0.000895128,
      "model_seconds": 22.38,
      "prompt_chars": 1273
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 45.04,
     "usage": {
      "prompt_tokens": 1782,
      "output_tokens": 659,
      "prompt_seconds": 12.630175,
      "output_seconds": 32.393002,
      "load_seconds": 0.001004364,
      "model_seconds": 45.04,
      "prompt_chars": 4886
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      }
     ],
     "passed": 6,
     "total": 6,
     "rejected": 2
    },
    {
     "scenario": "follow_up",
     "seconds": 9.47,
     "usage": {
      "prompt_tokens": 1010,
      "output_tokens": 109,
      "prompt_seconds": 4.373328,
      "output_seconds": 5.049041,
      "load_seconds": 0.001101036,
      "model_seconds": 9.47,
      "prompt_chars": 2015
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 19.2,
     "usage": {
      "prompt_tokens": 1230,
      "output_tokens": 277,
      "prompt_seconds": 6.523189,
      "output_seconds": 12.665209,
      "load_seconds": 0.000824471,
      "model_seconds": 19.2,
      "prompt_chars": 2944
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": false
      },
      {
       "label": "local costmap",
       "ok": false
      },
      {
       "label": "costmap color scheme",
       "ok": false
      },
      {
       "label": "footprint polygon",
       "ok": false
      }
     ],
     "passed": 1,
     "total": 5,
     "rejected": 2
    },
    {
     "scenario": "light_weight",
     "seconds": 20.32,
     "usage": {
      "prompt_tokens": 954,
      "output_tokens": 447,
      "prompt_seconds": 0.39442,
      "output_seconds": 19.893829,
      "load_seconds": 0.000944368,
      "model_seconds": 20.32,
      "prompt_chars": 1798
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 2
    },
    {
     "scenario": "camera",
     "seconds": 16.26,
     "usage": {
      "prompt_tokens": 734,
      "output_tokens": 323,
      "prompt_seconds": 1.864041,
      "output_seconds": 14.362891,
      "load_seconds": 0.000893571,
      "model_seconds": 16.26,
      "prompt_chars": 1069
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 16.11,
     "usage": {
      "prompt_tokens": 1720,
      "output_tokens": 94,
      "prompt_seconds": 11.515922,
      "output_seconds": 4.583491,
      "load_seconds": 0.000917956,
      "model_seconds": 16.11,
      "prompt_chars": 4533
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": false
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 6,
   "scenarios": 8,
   "checks_passed": 18,
   "checks": 23,
   "errors": 0,
   "rejected": 6,
   "median_seconds": 19.8,
   "mean_prompt_tokens": 1200,
   "mean_output_tokens": 315,
   "mean_prompt_seconds": 6.6,
   "mean_output_seconds": 14.6
  }
 ],
 "not_measured": [
  {
   "provider": "openai",
   "model": "gpt-5-mini",
   "reason": "API account had no credits"
  }
 ],
 "full_prompt_baseline": [
  {
   "provider": "ollama",
   "model": "qwen3:4b-instruct",
   "runs": [
    {
     "scenario": "fix_qos",
     "seconds": 56.02,
     "usage": {
      "prompt_tokens": 2301,
      "output_tokens": 334,
      "prompt_seconds": 26.729231,
      "output_seconds": 25.398063,
      "load_seconds": 3.846171246,
      "model_seconds": 56.02,
      "prompt_chars": 6761
     },
     "checks": [
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "cloud_height",
     "seconds": 45.52,
     "usage": {
      "prompt_tokens": 1854,
      "output_tokens": 424,
      "prompt_seconds": 14.582337,
      "output_seconds": 30.804926,
      "load_seconds": 0.00091337,
      "model_seconds": 45.52,
      "prompt_chars": 5070
     },
     "checks": [
      {
       "label": "PointCloud2 on /points",
       "ok": true
      },
      {
       "label": "AxisColor",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": true
      },
      {
       "label": "Best Effort for the lidar",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 4,
     "rejected": 0
    },
    {
     "scenario": "tf_robot",
     "seconds": 75.8,
     "usage": {
      "prompt_tokens": 2363,
      "output_tokens": 635,
      "prompt_seconds": 23.727695,
      "output_seconds": 51.923883,
      "load_seconds": 0.000955841,
      "model_seconds": 75.8,
      "prompt_chars": 7016
     },
     "checks": [
      {
       "label": "Fixed Frame \u2192 map",
       "ok": true
      },
      {
       "label": "RobotModel",
       "ok": true
      },
      {
       "label": "LaserScan \u2192 Best Effort",
       "ok": true
      },
      {
       "label": "Map \u2192 Transient Local",
       "ok": true
      },
      {
       "label": "PointCloud2",
       "ok": true
      },
      {
       "label": "Axis Z",
       "ok": false
      }
     ],
     "passed": 5,
     "total": 6,
     "rejected": 0
    },
    {
     "scenario": "follow_up",
     "seconds": 26.92,
     "usage": {
      "prompt_tokens": 2053,
      "output_tokens": 105,
      "prompt_seconds": 19.024225,
      "output_seconds": 7.688272,
      "load_seconds": 0.001026344,
      "model_seconds": 26.92,
      "prompt_chars": 5812
     },
     "checks": [
      {
       "label": "bigger point size on the cloud",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "nav2_view",
     "seconds": 90.81,
     "usage": {
      "prompt_tokens": 2408,
      "output_tokens": 869,
      "prompt_seconds": 23.14296,
      "output_seconds": 67.497687999,
      "load_seconds": 0.001499035,
      "model_seconds": 90.81,
      "prompt_chars": 7225
     },
     "checks": [
      {
       "label": "Path on /plan",
       "ok": true
      },
      {
       "label": "global costmap",
       "ok": true
      },
      {
       "label": "local costmap",
       "ok": true
      },
      {
       "label": "costmap color scheme",
       "ok": false
      },
      {
       "label": "footprint polygon",
       "ok": true
      }
     ],
     "passed": 4,
     "total": 5,
     "rejected": 0
    },
    {
     "scenario": "light_weight",
     "seconds": 51.38,
     "usage": {
      "prompt_tokens": 1997,
      "output_tokens": 697,
      "prompt_seconds": 0.736946,
      "output_seconds": 50.432773,
      "load_seconds": 0.001145507,
      "model_seconds": 51.38,
      "prompt_chars": 5595
     },
     "checks": [
      {
       "label": "a load-reducing setting",
       "ok": true
      },
      {
       "label": "says real downsampling needs a filter node",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    },
    {
     "scenario": "camera",
     "seconds": 46.96,
     "usage": {
      "prompt_tokens": 1877,
      "output_tokens": 425,
      "prompt_seconds": 16.732335,
      "output_seconds": 30.058826,
      "load_seconds": 0.000968587,
      "model_seconds": 46.96,
      "prompt_chars": 5169
     },
     "checks": [
      {
       "label": "Image on /front_camera/image_raw",
       "ok": true
      }
     ],
     "passed": 1,
     "total": 1,
     "rejected": 0
    },
    {
     "scenario": "hide_rename",
     "seconds": 35.9,
     "usage": {
      "prompt_tokens": 2302,
      "output_tokens": 174,
      "prompt_seconds": 22.912475,
      "output_seconds": 12.847398,
      "load_seconds": 0.001033929,
      "model_seconds": 35.9,
      "prompt_chars": 6767
     },
     "checks": [
      {
       "label": "Grid hidden",
       "ok": true
      },
      {
       "label": "LaserScan renamed",
       "ok": true
      }
     ],
     "passed": 2,
     "total": 2,
     "rejected": 0
    }
   ],
   "scenarios_passed": 6,
   "scenarios": 8,
   "checks_passed": 21,
   "checks": 23,
   "errors": 0,
   "rejected": 0,
   "median_seconds": 49.2,
   "mean_prompt_tokens": 2144,
   "mean_output_tokens": 458,
   "mean_prompt_seconds": 18.4,
   "mean_output_seconds": 34.6
  }
 ]
}