{
  "schema_version": 1,
  "recorded_at": "2026-07-27",
  "task": {
    "benchmark": "GeoAnalystBench",
    "id": 34,
    "title": "Estimate the accessibility of roads to rural areas in Japan",
    "source_url": "https://github.com/GeoDS/GeoAnalystBench",
    "authority": "Repository owner attested academic research and dissemination use on 2026-07-27",
    "input_publication": "private inputs; not uploaded to Arena"
  },
  "input_manifest": [
    {
      "path": "dataset/AllSeasonRoads.geojson",
      "bytes": 5483403,
      "sha256": "25d43c4e0a4dacba66862aa3b46644e767aded8f24d388f40f3bd7fdab40800c"
    },
    {
      "path": "dataset/ShikokuMetropolitan.geojson",
      "bytes": 800983,
      "sha256": "8c4e77d6c2bb087a268bff93bf3ea2f93a6ae49edf74bba040fd8133cc6180db"
    },
    {
      "path": "dataset/ShikokuPopulation.geojson",
      "bytes": 69615786,
      "sha256": "5ab3445c978d4ad37ac19d8e7540354e44e622656bc6432077989412f24fb937"
    }
  ],
  "reference": {
    "script_sha256": "fedda4d8e46af51538d8cffa745a2d7b3811f70ce3e0b4329fc91d93f8003781",
    "expected_png_sha256": "de9f38a9995cfeee15bc7c39fe062a6728a2cd09be3ecf3af4ce3240e66293dc",
    "rerun_png_sha256": "30fb33b5322f1a91ec591e8cc7a6b7c2e41fc7ae11fc01566c2130703008452d",
    "pixel_equal": true,
    "semantic_status": "incomplete_gold"
  },
  "oracles": {
    "published_tutorial": {
      "rural_population": 872986,
      "accessible_population": 736316,
      "access_percent_description": "about 84 percent"
    },
    "benchmark_compatible_epsg_4087": {
      "rural_population": 872986.0,
      "accessible_population": 736231.6135763831,
      "access_percent": 84.33487061377653
    },
    "spatial_validity_epsg_6672": {
      "rural_population": 872986.0,
      "accessible_population": 750557.1763424659,
      "access_percent": 85.97585486393436
    }
  },
  "generation": {
    "provider": "Cloudflare Workers AI",
    "models_attempted": 8,
    "prompt_contract": "task prompt plus exact PNG path and machine-readable JSON sidecar",
    "execution": "local bubblewrap sandbox with network namespace isolated"
  },
  "model_observations": [
    {
      "model": "@cf/moonshotai/kimi-k2.6",
      "attempts": 2,
      "terminal": "executed",
      "benchmark_match": true,
      "spatial_validity_match": false,
      "access_percent": 84.33487061377654
    },
    {
      "model": "@cf/zai-org/glm-4.7-flash",
      "attempts": 1,
      "terminal": "runtime_error",
      "failure": "GeoSeries has no dissolve method"
    },
    {
      "model": "@cf/zai-org/glm-5.2",
      "attempts": 1,
      "terminal": "executed",
      "benchmark_match": false,
      "spatial_validity_match": false,
      "access_percent": 86.21007911596725
    },
    {
      "model": "@cf/moonshotai/kimi-k2.7-code",
      "attempts": 1,
      "terminal": "executed",
      "benchmark_match": true,
      "spatial_validity_match": false,
      "access_percent": 84.33487061377653
    },
    {
      "model": "@cf/openai/gpt-oss-20b",
      "attempts": 1,
      "terminal": "executed_wrong_result",
      "access_percent": 78.94
    },
    {
      "model": "@cf/google/gemma-4-26b-a4b-it",
      "attempts": 2,
      "terminal": "no_visible_program",
      "failure": "output limit exhausted"
    },
    {
      "model": "@cf/qwen/qwen3-30b-a3b-fp8",
      "attempts": 2,
      "terminal": "runtime_error",
      "failure": "NumPy int64 was not JSON serializable"
    },
    {
      "model": "@cf/mistralai/mistral-small-3.1-24b-instruct",
      "attempts": 1,
      "terminal": "runtime_error",
      "failure": "municipality layer has no ID field"
    }
  ],
  "stored_artifacts": [
    {
      "model": "@cf/moonshotai/kimi-k2.6",
      "key_prefix": "research/task34/2026-07-27/workers-ai/kimi-k2.6/",
      "files": {
        "accessibility.png": "5f7e4a2a2c18e0a3318cf22e3ecb339a2ea1c7a5db587bab1bac8932eb5c7b22",
        "accessibility.json": "c1facbd72181e2602426e938eb149c41cd78b5903509f398585db172617c1a0c",
        "solution.py": "75212741c9739bc2f497ef61bc1f802a7b7bde828e1d74a47ad721c01fef91b5"
      },
      "http_readback_hash_match": true
    },
    {
      "model": "@cf/moonshotai/kimi-k2.7-code",
      "key_prefix": "research/task34/2026-07-27/workers-ai/kimi-k2.7-code/",
      "files": {
        "accessibility.png": "18df85388a4a6af64e6ebac0276ccdd20ffac57f0d190c7fd0539e25fb4b55e1",
        "accessibility.json": "9f2c7efb2974074c19d380ca5421532f8655c40bc520240753789f3484059bfc",
        "solution.py": "02f99037a1094d0341e43f5c7b50caded4eccf34754f066d6d7d1cc373c882bc"
      },
      "http_readback_hash_match": true
    }
  ],
  "publication": {
    "worker_version_id": "1d01d71f-275a-4449-8f97-7ec2d1a1546a",
    "page_url": "https://arena.axisspatial.com/evidence/task34",
    "api_url": "https://arena.axisspatial.com/api/evidence/task34",
    "source_inputs_served_by_arena": false,
    "live_page_claims": [
      "rights scope",
      "benchmark-gold validity",
      "actual agent performance"
    ]
  },
  "verdict": {
    "operating_path": "proved",
    "benchmark_gold": "reproducible_image_but_semantically_incomplete",
    "benchmark_match_models": [
      "@cf/moonshotai/kimi-k2.6",
      "@cf/moonshotai/kimi-k2.7-code"
    ],
    "accepted_spatial_validity_gold": null,
    "does_not_prove": [
      "model reliability",
      "pass rate",
      "full GeoAnalystBench reproduction",
      "production Worker and E2B path",
      "accepted spatially valid Task 34 gold"
    ]
  },
  "next_action": "Publish one Task 34 practitioner evidence card with separate benchmark-match and spatial-validity verdicts"
}