{
  "protocol": {
    "benchmark": "https://github.com/opper-ai/jevman-benchmark",
    "revision": "412ed72484ffeb3bd205a64be2d2ea5e233aac88",
    "map": "Original fixed map",
    "max_gameplay_seconds": 300,
    "parallel_games": 1,
    "added_latency_ms": 0,
    "local_hardware": "NVIDIA DGX Spark / GB10",
    "local_loops": 2,
    "local_probability_calibration": false,
    "local_server": "Fresh server per game; startup excluded from gameplay",
    "hosted_route": "OpenRouter; HTTP latency includes network and provider overhead",
    "sampling": "Repeated real-time games; not independent maps or paired random seeds",
    "jev_sampling": "All 12 completed games from two six-game runs using the same protocol",
    "replay_verification": "All completed replays passed the upstream verifier"
  },
  "conditions": [
    {
      "model": "Decisive Dory",
      "route": "NVFP4 \u00b7 Eider",
      "games": [
        {
          "score": 2300,
          "survived_seconds": 92.66160000000217,
          "calls": 191,
          "latency_ms_sum": 30752,
          "replay_sha256": "ab452665d336a32a32a02f92b1fbe2412afb0107f7d6a664d98a8a6ba8d5afaf"
        },
        {
          "score": 2040,
          "survived_seconds": 97.0343000000044,
          "calls": 212,
          "latency_ms_sum": 34994,
          "replay_sha256": "962baf1ab3627b119e0884158bbb7eb9f47285ea3f40c49179f7cdf376239a22"
        },
        {
          "score": 2830,
          "survived_seconds": 96.17950000000396,
          "calls": 195,
          "latency_ms_sum": 30941,
          "replay_sha256": "77f0471014575ad5c2f27d9878d71297300a9d5038562951f5f7047526fd43e7"
        },
        {
          "score": 2260,
          "survived_seconds": 76.8986000000027,
          "calls": 150,
          "latency_ms_sum": 24841,
          "replay_sha256": "226832364e02185418f46e0b2ae7051fe692f2eb51283c72e2322f42d9a09c15"
        },
        {
          "score": 2690,
          "survived_seconds": 81.84360000000173,
          "calls": 159,
          "latency_ms_sum": 24517,
          "replay_sha256": "6f58daea7fa64b0eb14520b820a30ff6fa1ab21b89add0a31649f3e36899b332"
        },
        {
          "score": 1910,
          "survived_seconds": 67.23850000000128,
          "calls": 143,
          "latency_ms_sum": 25136,
          "replay_sha256": "fd5b20647dd61a04aa71565e5540e3ac0b8d2f25a8878a742790f3af33271982"
        }
      ],
      "game_count": 6,
      "mean_score": 2338.3333333333335,
      "median_score": 2280.0,
      "mean_decision_ms": 163.02952380952382,
      "scheduler_errors": 0,
      "resolved_models": [
        "local-dory"
      ]
    },
    {
      "model": "Decisive Dory",
      "route": "Q8_0 \u00b7 llama.cpp",
      "games": [
        {
          "score": 2220,
          "survived_seconds": 90.00120000000175,
          "calls": 151,
          "latency_ms_sum": 48905,
          "replay_sha256": "97f4bcc396b43e9e8848bd0093977b531aa8193c1871d3f26917c0ea290acb39"
        },
        {
          "score": 2310,
          "survived_seconds": 65.34389999999954,
          "calls": 114,
          "latency_ms_sum": 40117,
          "replay_sha256": "1e9b0d7114f5fa33d5719e8d20bfb89ffdc96d2753e4a128ee31e94797f6ca8f"
        },
        {
          "score": 2310,
          "survived_seconds": 64.91600000000018,
          "calls": 114,
          "latency_ms_sum": 40300,
          "replay_sha256": "efc1ee7b3c96469a866e5aadd94708104a76dbc6520f69dc3247a32a03af412f"
        },
        {
          "score": 2130,
          "survived_seconds": 58.775600000001006,
          "calls": 102,
          "latency_ms_sum": 35172,
          "replay_sha256": "c2ddfdb40daf0a7ca943d6c012690bef8b5ac49f9589681c7e09e9483ad0d002"
        },
        {
          "score": 2220,
          "survived_seconds": 90.04169999999984,
          "calls": 153,
          "latency_ms_sum": 51135,
          "replay_sha256": "c0b76b482c1aabec20cd11b314051537517fedc3012f998995d9a3f0bf0fcd6d"
        },
        {
          "score": 2980,
          "survived_seconds": 87.8558000000031,
          "calls": 152,
          "latency_ms_sum": 52981,
          "replay_sha256": "287970b9664eea214c9c28dd092d1a6dc942430eabc8394a2970360bb357fce5"
        }
      ],
      "game_count": 6,
      "mean_score": 2361.6666666666665,
      "median_score": 2265.0,
      "mean_decision_ms": 341.74300254452925,
      "scheduler_errors": 0,
      "resolved_models": [
        "dory-jevman-q8"
      ]
    },
    {
      "model": "Jev 1.14",
      "route": "OpenRouter",
      "games": [
        {
          "score": 4770,
          "survived_seconds": 66.7314000000021,
          "calls": 135,
          "latency_ms_sum": 33957,
          "replay_sha256": "f33346b8d20393805575649c416a2c4343e9d7c4e6011009188dc5c3b055e662"
        },
        {
          "score": 2460,
          "survived_seconds": 46.36650000000062,
          "calls": 95,
          "latency_ms_sum": 16876,
          "replay_sha256": "2a295eb5464d7ce30be8323147cd936ce40a85473f3a110e7ff5c5c2960827ce"
        },
        {
          "score": 2540,
          "survived_seconds": 45.76800000000065,
          "calls": 93,
          "latency_ms_sum": 17012,
          "replay_sha256": "f8d9967634a0b6a067e158213785c57e51af9ad2763b457e925a28d360110ea5"
        },
        {
          "score": 6050,
          "survived_seconds": 105.78640000000006,
          "calls": 204,
          "latency_ms_sum": 35635,
          "replay_sha256": "b299862de3623ebb332c36646d3d202cda701b0d02495dba118141eb050c12d4"
        },
        {
          "score": 3020,
          "survived_seconds": 65.42200000000238,
          "calls": 124,
          "latency_ms_sum": 22516,
          "replay_sha256": "817ef7cf8f250fcde6d8ef31ac5c511977cc7822e4c80938fbc5a7f9a607e367"
        },
        {
          "score": 5170,
          "survived_seconds": 113.07670000000334,
          "calls": 211,
          "latency_ms_sum": 41874,
          "replay_sha256": "dca22fa9a3b7e7e7adf20fb7350ed04f5b75869e0c7eebecfb9cdef00dc8471e"
        },
        {
          "score": 3360,
          "survived_seconds": 65.91620000000175,
          "calls": 115,
          "latency_ms_sum": 27609,
          "replay_sha256": "f3f2a114c9c196317d76daad98ffeb7760f6ebfeedaec2910369274670d9fcb0"
        },
        {
          "score": 2510,
          "survived_seconds": 56.38020000000047,
          "calls": 110,
          "latency_ms_sum": 20643,
          "replay_sha256": "405afa6aa1e283518d90e4b4e2105be5e79e9c726d2c9d758431e4d0226c231c"
        },
        {
          "score": 6610,
          "survived_seconds": 100.2825000000033,
          "calls": 196,
          "latency_ms_sum": 33076,
          "replay_sha256": "ee10c68d2ef0f7081379508d658bc45ab926cb86a36d5be53105d249ac8f9026"
        },
        {
          "score": 2400,
          "survived_seconds": 50.01130000000068,
          "calls": 95,
          "latency_ms_sum": 23380,
          "replay_sha256": "425ff1f1258f15bad585284621e1ae135d6ab50092b0fedace4915bc9a945365"
        },
        {
          "score": 2190,
          "survived_seconds": 50.25400000000027,
          "calls": 88,
          "latency_ms_sum": 22038,
          "replay_sha256": "49f635dabfa938af30377099bdaa9e43666402997c8a08153b34ce31d570d325"
        },
        {
          "score": 3370,
          "survived_seconds": 78.94970000000323,
          "calls": 138,
          "latency_ms_sum": 29063,
          "replay_sha256": "3cb6f1436d33144467147d9e4ec45fc8d4ce23eb43d571ba3e3cbfc8762ace4a"
        }
      ],
      "game_count": 12,
      "mean_score": 3704.1666666666665,
      "median_score": 3190.0,
      "mean_decision_ms": 201.79488778054863,
      "scheduler_errors": 4,
      "resolved_models": [
        "typesafe/jev-1.14-20261014"
      ]
    }
  ],
  "eider_weights_sha256": "62bc20a6fbafaf157154650dfb4e9ec32a19d67e8e1f60ddcc3318fd37d6ccb4",
  "llama_q8_weights_sha256": "bb36b8ed263ce633649ccfad9c6e1dc065e9c3d333624fc27028f496089a38b8",
  "base_reference": {
    "model": "NVIDIA Dory 1.0 4B (unadapted)",
    "route": "NVFP4 \u00b7 Eider",
    "games": [
      {
        "score": 1490,
        "survived_seconds": 48.98480000000033,
        "calls": 115,
        "latency_ms_sum": 22525,
        "replay_sha256": "f06d720ed6fa6881a82cbb58ab0da4446e9553ee620b3d82047872e52577a27c"
      },
      {
        "score": 680,
        "survived_seconds": 35.804200000000534,
        "calls": 82,
        "latency_ms_sum": 14881,
        "replay_sha256": "6eae5fd6214d64446a285712a0e4b76a736edf90b9ea78ebedce31fdb3b1b8c1"
      },
      {
        "score": 1210,
        "survived_seconds": 36.868400000000015,
        "calls": 83,
        "latency_ms_sum": 13773,
        "replay_sha256": "64a36d3e4c9d8c4be8e20caac33c39478f84ca81190151d9538022ef3e744935"
      },
      {
        "score": 1080,
        "survived_seconds": 32.14370000000079,
        "calls": 62,
        "latency_ms_sum": 10051,
        "replay_sha256": "4090ddef29ed6b35294af7153ae6eb46de68a7c2ec60f381c1b5b1ec279c97e7"
      },
      {
        "score": 1490,
        "survived_seconds": 42.16530000000059,
        "calls": 90,
        "latency_ms_sum": 14641,
        "replay_sha256": "155a735437f3788abbddee63382697b0b7b0d47ba4ea0aa77e5f07a2f13254d0"
      },
      {
        "score": 630,
        "survived_seconds": 33.85650000000001,
        "calls": 75,
        "latency_ms_sum": 13615,
        "replay_sha256": "32499a938cc6265eae1f45f313d5962bdb79d8a07e14ee5bcec4d45766e045ea"
      }
    ],
    "game_count": 6,
    "mean_score": 1096.6666666666667,
    "median_score": 1145.0,
    "mean_decision_ms": 176.5009861932939,
    "scheduler_errors": 0,
    "weights_sha256": "2dc8446f2c52d3249eac5bd2154f1f4a8f206152e3b66d7077965b9ad991bbf7",
    "quantization": "Same ModelOpt recipe and calibration examples as Decisive Dory"
  }
}