{
  "schema": "synth.evals.gamebench.media.v1",
  "generated_at": "2026-07-30",
  "gamebench_revision": "05b295e07960282d8a49b35d8a4e1b68db19c63a",
  "policy": "Only media captured from an executed GameBench code policy may appear as rollout evidence. Environment art, fixtures, and illustrative mockups are not rollout evidence.",
  "assets": [
    {
      "id": "emerald-bedroom-explorer-seed1",
      "environment": "pokemon-emerald-littleroot-singleplayer",
      "candidate": "bedroom_explorer_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/emerald-bedroom-explorer-seed1.webp",
      "trace": "traces/emerald-bedroom-explorer-seed1.json",
      "summary": "traces/emerald-policy-dev-v1-summary.json",
      "policy_snapshot": "policies/emerald/bedroom-explorer-v1.py",
      "checkpoint": "bedroom_idle",
      "seed": 1,
      "decisions": 8,
      "emulated_frames": 128,
      "captured_frames": 9,
      "suite": "emerald_policy_dev_v1",
      "publication_status": "superseded",
      "publication_note": "This capture and seven-scenario summary remain as historical execution evidence only. The current local emerald_policy_dev_v1 suite contains fourteen scenarios, so its reported 0.2857 must not be presented as a current or comparable benchmark result until the suite and artifacts are rerun together.",
      "provenance_note": "The policy and sweep files named by this artifact were local uncommitted files and are not present at gamebench_revision. The included policy snapshot and listed SHA-256 digests identify the captured bytes, but the revision does not independently reproduce the run.",
      "result": {
        "score": 0.2857,
        "success_rate": 0.2857,
        "mean_reward": 2.4286,
        "scenarios": 7
      },
      "sha256": {
        "rollout": "058a18a48006d19c9f9adea761d092c450a06add03b8b76dcde213fed2c07179",
        "trace": "7a1611e5243f09c983d72cd5723edc01296d9606f54bb08aad035473eebcc2f6",
        "summary": "2c52ac20e521c2989ff9b02cc9c7cdb4751c9b8c9469bfc66c02eb45ab31845",
        "policy_snapshot": "1a16f22adec279ed4c3d665ab5798fe6994814c7ff25c9c305ec365c6844e5f1"
      }
    },
    {
      "id": "craftax-mp-coordination-all-do-seed101",
      "environment": "craftax-multiplayer",
      "candidate": "coordination_all_do_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/craftax-mp/coordination-all-do-seed101.webp",
      "trace": "traces/craftax-mp-coordination-all-do-seed101.json",
      "summary": "traces/craftax-mp-policy-dev-summary.json",
      "policy_snapshot": "policies/craftax-mp/coordination-all-do-v1.py",
      "scenario": "sync_2",
      "alpha": 0.3,
      "seed": 101,
      "captured_frames": 4,
      "suite": "craftax_coop_alem_coord_deo_v1",
      "result": {
        "score": 0.7,
        "coord_success_rate": 0.6666666666666666,
        "representative_rollout_success": true,
        "representative_coord_reward": 2.0,
        "baseline_score": 0.1
      },
      "sha256": {
        "rollout": "e03cfdfac9578181279f29aa9d721599656c148f24d82c1e967a3e4918b21d2e",
        "trace": "f5a461d7e2d4c0ce2fbc614c969dc0af49377b548db680c125cf726257afa570",
        "summary": "68733d7f82314d5a5dfd0aae974d778822df8d18422d2698ce23c3b93cf45e8d",
        "policy_snapshot": "34aa482d136c18e0107b54c6c5c434d45fcb8852c997eaebce76d54368c65ab3"
      }
    },
    {
      "id": "dungeongrid-sp-lantern-crypt",
      "environment": "dungeongrid-singleplayer",
      "candidate": "heuristic_baseline",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/dungeongrid-sp/heuristic-baseline-lantern-crypt.webp",
      "trace": "traces/dungeongrid-sp-lantern-crypt.json",
      "summary": "traces/dungeongrid-sp-policy-dev-summary.json",
      "policy_snapshot": "policies/dungeongrid-sp/heuristic-baseline.py",
      "scenario": "lantern_crypt_lite",
      "captured_frames": 8,
      "suite": "dungeongrid_singleplayer_policy_dev_v1",
      "result": {
        "score": 0.9,
        "success_rate": 0.0,
        "representative_reward": 0.6,
        "representative_achievements": [
          "coordination.guard_used",
          "coordination.message_sent",
          "movement.first_step",
          "routing.opened_door"
        ]
      },
      "sha256": {
        "rollout": "7b771fecb9d222071f9d8e92c5fde0cb735344a51a89f36f5ad3a986d555b5b2",
        "trace": "00412b38dada446b175ab86fe9821f9506f6be3c7e1c7796726b4495400040a9",
        "summary": "b1e4ff7c99b3b78c0428834c3478cbd31fa6f19367ad49b82f86fe88fe99b052",
        "policy_snapshot": "19794eb8bcb72a43068d3f916f93da99ffa2598225f009e5b4fa3f95fc610434"
      }
    },
    {
      "id": "overcooked-bfs-joint-seed8",
      "environment": "overcooked-v2-multiplayer",
      "candidate": "bfs_joint_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/overcooked/bfs-joint-seed8.webp",
      "trace": "traces/overcooked-bfs-joint-seed8.json",
      "summary": "traces/overcooked-policy-dev-summary.json",
      "policy_snapshot": "policies/overcooked/bfs-joint-v1.py",
      "scenario": "policy_dev_delivery",
      "seed": 8,
      "captured_frames": 18,
      "suite": "overcooked_v2_policy_dev_v1",
      "result": {
        "score": 0.7778,
        "success_rate": 0.7778,
        "representative_deliveries": 1,
        "representative_reward": 1.0,
        "representative_rollout_success": true
      },
      "sha256": {
        "rollout": "9f4b42df7d551506d5bbe83d07db961b294ea3e8f7a80e86b86f19c50be510f8",
        "trace": "1d599a0411fc2e81cb0c2e8118025a43b559992a28c80bdb852139fba14672d7",
        "summary": "b52a8e93fd4ede6321a2fc78c801755006b8e60b71d03e60f8bbdcfde5453f2e",
        "policy_snapshot": "c78d1c38aae38c14de550615d1c513a8de6a74df02573ccab30f6307a028e81b"
      }
    },
    {
      "id": "sokoban-greedy-seed103",
      "environment": "sokoban-singleplayer",
      "candidate": "greedy_distance_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/sokoban/greedy-seed103.webp",
      "trace": "traces/sokoban-greedy-seed103.json",
      "summary": "traces/sokoban-policy-dev-summary.json",
      "policy_snapshot": "policies/sokoban/greedy-distance-v1.py",
      "policy_dependency": "policies/sokoban/registry.py",
      "scenario": "seed-109",
      "seed": 103,
      "steps": 120,
      "rendered_samples": 33,
      "animated_pages": 4,
      "visual_renderer": {
        "name": "gym-sokoban public sprite renderer",
        "repository": "https://github.com/mpSchrader/gym-sokoban",
        "revision": "8e06e44e8bf3bb8bc73eeb1e7f0354508ce3fc89",
        "license": "MIT",
        "source_manifest": "third-party/gym-sokoban/SOURCE.json",
        "license_file": "third-party/gym-sokoban/LICENSE",
        "semantic_boundary": "Only the visual tiles come from gym-sokoban. GameBench remains authoritative for the policy, state transitions, trace, and score."
      },
      "suite": "policy_dev_v1",
      "result": {
        "score": 0.155,
        "success_rate": 0.1,
        "mean_reward": -3.0569999999999924,
        "representative_reward": -3.4199999999999924,
        "representative_rollout_status": "truncated",
        "representative_achievements": [
          "first_push",
          "first_box_on_target"
        ]
      },
      "sha256": {
        "rollout": "28f91e4b7ab78fc7d78d6bf2d15b2ec0ee910292562c7d1e95dd465602ce9c14",
        "trace": "ac7d451f862fc562d9071c4c3237d881893574f07839cbc5d143b604b4d38483",
        "summary": "cfdbce8ffbea581366d1afb1018383c7412b5376d832054dbec7a1842af456e9",
        "policy_snapshot": "05192ba85ea3f7e401ea2e80576b9cc38d75317309597fbf825f834fe79c1b33",
        "policy_dependency": "2071db2434c96c95de8846764214c6a93d3f648e4bb98971d2689858b19f66a1"
      }
    },
    {
      "id": "craftax-baseline",
      "environment": "craftax-singleplayer",
      "candidate": "baseline",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/craftax/baseline.gif",
      "policy_snapshot": "policies/craftax/baseline.py",
      "source_origin": "/blog/gamebench/candidates/craftax/standalone/baseline.gif",
      "suite": "craftax_policy_dev_v1",
      "result": {
        "score": 0.0848,
        "mean_reward": 4.78,
        "seeds": 10
      },
      "sha256": "3c15067688763b65e49da9549a6da8d4014c6062a4d0afb404986a7b8e435ca6"
    },
    {
      "id": "craftax-hillclimb-v1",
      "environment": "craftax-singleplayer",
      "candidate": "hillclimb_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/craftax/hillclimb-v1.gif",
      "policy_snapshot": "policies/craftax/hillclimb-v1.py",
      "source_origin": "/blog/gamebench/candidates/craftax/smr/hillclimb_v1.gif",
      "suite": "craftax_policy_dev_v1",
      "result": {
        "score": 0.1515,
        "mean_reward": 9.19,
        "seeds": 10
      },
      "sha256": "0c0dd4e6c0f94370631cee6e50545e220e832694c5799c7b12a58e77636c27c1"
    },
    {
      "id": "craftax-attempt4",
      "environment": "craftax-singleplayer",
      "candidate": "attempt4",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/craftax/attempt4.gif",
      "policy_snapshot": "policies/craftax/attempt4.py",
      "source_origin": "/blog/gamebench/candidates/craftax/standalone/attempt4.gif",
      "suite": "craftax_policy_dev_v1",
      "result": {
        "score": 0.103,
        "mean_reward": 6.27,
        "seeds": 10
      },
      "sha256": "fa65ee50469232d0e0e566e2ad8b0fa440683e335a45e0178ec1f41681957d32"
    },
    {
      "id": "rogue-baseline",
      "environment": "rogue-singleplayer",
      "candidate": "baseline",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/rogue/baseline.gif",
      "policy_snapshot": "policies/rogue/baseline.py",
      "source_origin": "/blog/gamebench/candidates/rogue/smr/baseline.gif",
      "suite": "rogue_policy_dev_v1",
      "result": {
        "score": 1.0,
        "mean_scout_score": 55.0,
        "seeds": 4
      },
      "sha256": "9d0dcd5291bbce265fd96a8c24f6b204f09d6675ab636c8bdc065fdee502ff23"
    },
    {
      "id": "rogue-scout-detour-v1",
      "environment": "rogue-singleplayer",
      "candidate": "scout_detour_v1",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/rogue/scout-detour-v1.gif",
      "policy_snapshot": "policies/rogue/scout-detour-v1.py",
      "source_origin": "/blog/gamebench/candidates/rogue/smr/scout_detour_v1.gif",
      "suite": "rogue_policy_dev_v1",
      "result": {
        "score": 1.0,
        "mean_scout_score": 65.5,
        "seeds": 4
      },
      "sha256": "7ad3215b026533df353674098b9cd5a8b2f9c0e235552632da28e145c9e18ff6"
    },
    {
      "id": "rogue-door-blind-seed101",
      "environment": "rogue-singleplayer",
      "candidate": "door_blind_policy_puzzle",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/rogue/door-blind-seed101.gif",
      "policy_snapshot": "policies/rogue/puzzle-door-blind-v1.json",
      "policy_source": "tasks/rogue-singleplayer/policy_puzzles/tasks/puzzle_door_blind_v1.json",
      "source_origin": "/blog/gamebench/rollouts/rogue/puzzles/door-blind-seed101.gif",
      "seed": 101,
      "steps": 80,
      "sha256": {
        "rollout": "a92e6329d9a1227d9e60cf77f99772de86c43686f9749b5eb33694a390ad0475",
        "policy_snapshot": "e37cc29efbc25fd5481972c08fd651de02e35fdf7021aad1854c9c700fd6d819"
      }
    },
    {
      "id": "dungeongrid-top-policy-seed101",
      "environment": "dungeongrid-multiplayer",
      "candidate": "heuristic_baseline",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/dungeongrid/top-policy-seed101.gif",
      "policy_snapshot": "policies/dungeongrid-mp/heuristic-baseline.py",
      "summary": "traces/dungeongrid-mp-policy-dev-summary.json",
      "source_origin": "/blog/gamebench/rollouts/dungeongrid/deo/top-policy-seed101.gif",
      "seed": 101,
      "steps": 2,
      "suite": "dungeongrid_policy_dev_v1",
      "result": {
        "representative_rollout_status": "complete",
        "suite_score": 0.169232,
        "suite_success_rate": 0.0,
        "suite_mean_reward": 0.2
      },
      "sha256": {
        "rollout": "04ca519b26b69f94f698adccf93ed1f35bf2e30e0804ee07757a5e307c562616",
        "summary": "d3e515f350c2cc30bc4ec97a63a15b5da934e0495b85fa975d2c26b886589e4f",
        "policy_snapshot": "19794eb8bcb72a43068d3f916f93da99ffa2598225f009e5b4fa3f95fc610434"
      }
    },
    {
      "id": "dungeongrid-sparse-governor-seed101",
      "environment": "dungeongrid-multiplayer",
      "candidate": "sparse_governor",
      "kind": "published_cybernetic_rollout",
      "rollout": "rollouts/dungeongrid/sparse-governor-seed101.gif",
      "policy_source": "tasks/dungeongrid-multiplayer/exotic_cybernetics/reference/sparse_governor/heuristic_policy.py",
      "source_manifest": "/blog/gamebench/rollouts/render-manifest.json",
      "source_origin": "/blog/gamebench/rollouts/dungeongrid/cybernetic/sparse-governor-seed101.gif",
      "seed": 101,
      "steps": 2,
      "evidence_limit": "The historical policy source is absent from the current checkout; only the published render metadata is asserted.",
      "sha256": "04ca519b26b69f94f698adccf93ed1f35bf2e30e0804ee07757a5e307c562616"
    },
    {
      "id": "dungeongrid-protocol-missing-seed101",
      "environment": "dungeongrid-multiplayer",
      "candidate": "protocol_missing_policy_puzzle",
      "kind": "executed_code_policy_rollout",
      "rollout": "rollouts/dungeongrid/protocol-missing-seed101.gif",
      "policy_snapshot": "policies/dungeongrid-mp/puzzle-protocol-missing-v1.json",
      "policy_source": "tasks/dungeongrid-multiplayer/defaults/policy_sweep/policy_puzzle_protocol_missing_v1.json",
      "source_origin": "/blog/gamebench/rollouts/dungeongrid/puzzles/protocol-missing-seed101.gif",
      "seed": 101,
      "steps": 7,
      "sha256": {
        "rollout": "063d01f6390944b248dfbcde94149e4748d7cfb37e331552c8941952eda5ddff",
        "policy_snapshot": "f7c9dfb6f414214c6732741a4c601b7b474522b6331d882ca5ca7bf1d5fad9f6"
      }
    },
    {
      "id": "craftax-hunger-blind-seed101",
      "environment": "craftax-singleplayer",
      "candidate": "hunger_blind_policy_puzzle",
      "kind": "published_policy_puzzle_rollout",
      "rollout": "rollouts/craftax/hunger-blind-seed101.gif",
      "policy_source": "tasks/craftax-singleplayer/policy_puzzles/tasks/puzzle_hunger_blind_v1.json",
      "source_origin": "/blog/gamebench/rollouts/craftax/puzzles/hunger-blind-seed101.gif",
      "seed": 101,
      "steps": 49,
      "evidence_limit": "The historical rollout manifest names the policy source, but that source is absent from the current checkout.",
      "sha256": "e272022a0bb5b3504a15488b69493faec4f6a84d3fcca716c593194331741966"
    },
    {
      "id": "craftax-ai-planner-seed101",
      "environment": "craftax-singleplayer",
      "candidate": "ai_planner_cybernetic",
      "kind": "published_cybernetic_rollout",
      "rollout": "rollouts/craftax/ai-planner-seed101.gif",
      "source_manifest": "/blog/gamebench/rollouts/render-manifest.json",
      "source_origin": "/blog/gamebench/rollouts/craftax/cybernetic/ai-planner-seed101.gif",
      "seed": 101,
      "steps": 60,
      "evidence_limit": "The published render manifest records frames and steps but does not name a policy source.",
      "sha256": "8cbb416be53ca9cecfb9428b3b36945572a2561e3d8ab85a2c4ede4d9a17e793"
    },
    {
      "id": "rogue-sparse-governor-seed101",
      "environment": "rogue-singleplayer",
      "candidate": "sparse_governor",
      "kind": "published_cybernetic_rollout",
      "rollout": "rollouts/rogue/sparse-governor-seed101.gif",
      "policy_snapshot": "policies/rogue/sparse-governor.py",
      "policy_source": "tasks/rogue-singleplayer/exotic_cybernetics/reference/sparse_governor/heuristic_policy.py",
      "source_origin": "/blog/gamebench/rollouts/rogue/cybernetic/sparse-governor-seed101.gif",
      "seed": 101,
      "steps": 10,
      "sha256": {
        "rollout": "9d0dcd5291bbce265fd96a8c24f6b204f09d6675ab636c8bdc065fdee502ff23",
        "policy_snapshot": "beb31988c72e297ed29752f94ea519ce5d72da18a2f6877dabab06e66bd1ce2f"
      }
    }
  ],
  "coverage": {
    "verified_rollout_media": [
      "pokemon-emerald-littleroot-singleplayer",
      "craftax-singleplayer",
      "craftax-multiplayer",
      "rogue-singleplayer",
      "sokoban-singleplayer",
      "dungeongrid-singleplayer",
      "dungeongrid-multiplayer",
      "overcooked-v2-multiplayer"
    ],
    "aggregate_only_until_rollout_media_is_captured": [
      "minihack-singleplayer"
    ],
    "incoming_cross_language_results": [
      "harbor-python-to-rust",
      "dock-python-to-rust"
    ]
  }
}
