{
  "benchmark": {
    "name": "Terminal-Bench-Science",
    "score": "resolution-rate",
    "scoreUnit": "percent",
    "taskCount": 70,
    "trialsPerConfiguration": 210,
    "trialsPerTask": 3,
    "version": "0.1.0",
    "runPolicy": {
      "actionLimit": null,
      "agentCount": null,
      "agentTimeoutSeconds": null,
      "costAggregationPolicy": "preserve-owner-aggregate-and-domain-costs-without-reconciliation",
      "costBasis": "source-reported-total-evaluation-usd; pricing basis unspecified",
      "errorTreatment": null,
      "retryPolicy": null,
      "seedPolicy": null,
      "tokenLimit": null,
      "toolsMode": null,
      "uncertaintyMethod": "source-reported-binomial-standard-error"
    }
  },
  "records": [
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "76867ebc-8f35-43c5-a8b0-0cee9db46c96",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 19,
            "resolutionRatePercent": 79.16666666666666,
            "standardErrorPercent": 8.28981693493981,
            "totalCostUsd": 595.7791399999999,
            "totalTokens": 293274673
          },
          "engineering": {
            "nTrials": 27,
            "passes": 19,
            "resolutionRatePercent": 70.37037037037037,
            "standardErrorPercent": 8.787718725951697,
            "totalCostUsd": 698.2211565000001,
            "totalTokens": 275452102
          },
          "life": {
            "nTrials": 57,
            "passes": 34,
            "resolutionRatePercent": 59.64912280701754,
            "standardErrorPercent": 6.498170362001764,
            "totalCostUsd": 1029.8967839999998,
            "totalTokens": 499877944
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 41,
            "resolutionRatePercent": 80.3921568627451,
            "standardErrorPercent": 5.559511900620688,
            "totalCostUsd": 1453.6847160000002,
            "totalTokens": 661648841
          },
          "physical": {
            "nTrials": 51,
            "passes": 30,
            "resolutionRatePercent": 58.82352941176471,
            "standardErrorPercent": 6.891519836811836,
            "totalCostUsd": 1219.8793844999996,
            "totalTokens": 558073236
          }
        },
        "nTrials": 210,
        "passes": 143,
        "resolutionRatePercent": 68.0952380952381,
        "standardErrorPercent": 3.2164475807033743,
        "totalCostUsd": 4997.461181,
        "totalTokens": 2288326796
      },
      "model": {
        "display": {
          "label": "GPT-6 Astra",
          "url": "https://developers.openai.com/api/docs/models/gpt-6-astra"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "rank": 1,
      "reasoningEffort": "max",
      "releaseDate": "2026-09-03",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-22T11:46:04.426644+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/76867ebc-8f35-43c5-a8b0-0cee9db46c96"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "cb24f3a2-82ab-4f97-be88-92a28118bf1c",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 15,
            "resolutionRatePercent": 62.5,
            "standardErrorPercent": 9.882117688026186,
            "totalCostUsd": 638.2260182000001,
            "totalTokens": 1340109649
          },
          "engineering": {
            "nTrials": 27,
            "passes": 17,
            "resolutionRatePercent": 62.96296296296296,
            "standardErrorPercent": 9.293489539618284,
            "totalCostUsd": 1028.4841072,
            "totalTokens": 1652861078
          },
          "life": {
            "nTrials": 57,
            "passes": 31,
            "resolutionRatePercent": 54.385964912280706,
            "standardErrorPercent": 6.597132965516611,
            "totalCostUsd": 970.3947527000001,
            "totalTokens": 1929930990
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 39,
            "resolutionRatePercent": 76.47058823529412,
            "standardErrorPercent": 5.939743109381192,
            "totalCostUsd": 1159.0756179999998,
            "totalTokens": 1684896896
          },
          "physical": {
            "nTrials": 51,
            "passes": 31,
            "resolutionRatePercent": 60.78431372549019,
            "standardErrorPercent": 6.836606453024293,
            "totalCostUsd": 1078.0771736999998,
            "totalTokens": 2139129551
          }
        },
        "nTrials": 210,
        "passes": 133,
        "resolutionRatePercent": 63.33333333333333,
        "standardErrorPercent": 3.3253873546101445,
        "totalCostUsd": 4874.2576698,
        "totalTokens": 8746928164
      },
      "model": {
        "display": {
          "label": "Opus 5.5",
          "url": "https://www.anthropic.com/claude-opus-5-5"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "rank": 2,
      "reasoningEffort": "max",
      "releaseDate": "2026-09-22",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-23T16:29:34.906395+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/cb24f3a2-82ab-4f97-be88-92a28118bf1c"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "1264a994-632f-4742-a143-a848a6ad4114",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 9,
            "resolutionRatePercent": 37.5,
            "standardErrorPercent": 9.882117688026186,
            "totalCostUsd": 746.4760759999999,
            "totalTokens": 431225074
          },
          "engineering": {
            "nTrials": 27,
            "passes": 13,
            "resolutionRatePercent": 48.148148148148145,
            "standardErrorPercent": 9.615902424319279,
            "totalCostUsd": 976.645512,
            "totalTokens": 633357580
          },
          "life": {
            "nTrials": 57,
            "passes": 22,
            "resolutionRatePercent": 38.59649122807017,
            "standardErrorPercent": 6.448119524174796,
            "totalCostUsd": 1109.489021,
            "totalTokens": 751380608
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 21,
            "resolutionRatePercent": 41.17647058823529,
            "standardErrorPercent": 6.891519836811836,
            "totalCostUsd": 1596.16417625,
            "totalTokens": 891442338
          },
          "physical": {
            "nTrials": 51,
            "passes": 19,
            "resolutionRatePercent": 37.254901960784316,
            "standardErrorPercent": 6.770122476910462,
            "totalCostUsd": 1818.3650532499998,
            "totalTokens": 1045886930
          }
        },
        "nTrials": 210,
        "passes": 84,
        "resolutionRatePercent": 40,
        "standardErrorPercent": 3.380617018914066,
        "totalCostUsd": 6247.1398385,
        "totalTokens": 3753292530
      },
      "model": {
        "display": {
          "label": "Fable 5.1",
          "url": "https://www.anthropic.com/claude-fable-and-mythos-5-1"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "rank": 3,
      "reasoningEffort": "max",
      "releaseDate": "2026-09-01",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/1264a994-632f-4742-a143-a848a6ad4114"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "bcf50bd8-8555-48a6-8cae-9f697af54051",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 11,
            "resolutionRatePercent": 45.83333333333333,
            "standardErrorPercent": 10.170707302692229,
            "totalCostUsd": 1150.81077375,
            "totalTokens": 897476528
          },
          "engineering": {
            "nTrials": 27,
            "passes": 8,
            "resolutionRatePercent": 29.629629629629626,
            "standardErrorPercent": 8.787718725951697,
            "totalCostUsd": 1035.3482442499999,
            "totalTokens": 1225811096
          },
          "life": {
            "nTrials": 57,
            "passes": 17,
            "resolutionRatePercent": 29.82456140350877,
            "standardErrorPercent": 6.059575107331345,
            "totalCostUsd": 1324.0133410000003,
            "totalTokens": 1306678199
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 13,
            "resolutionRatePercent": 25.49019607843137,
            "standardErrorPercent": 6.102505932917984,
            "totalCostUsd": 1634.1868124999999,
            "totalTokens": 1709641761
          },
          "physical": {
            "nTrials": 51,
            "passes": 14,
            "resolutionRatePercent": 27.450980392156865,
            "standardErrorPercent": 6.2489869213841,
            "totalCostUsd": 1848.3181522499997,
            "totalTokens": 2127060729
          }
        },
        "nTrials": 210,
        "passes": 63,
        "resolutionRatePercent": 30,
        "standardErrorPercent": 3.162277660168379,
        "totalCostUsd": 6992.67732375,
        "totalTokens": 7266668313
      },
      "model": {
        "display": {
          "label": "Opus 5",
          "url": "https://www.anthropic.com/news/claude-opus-5"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "rank": 4,
      "reasoningEffort": "max",
      "releaseDate": "2026-07-24",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/bcf50bd8-8555-48a6-8cae-9f697af54051"
    },
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "378f8b49-dbd8-4acf-a065-8c07e1712de2",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 5,
            "resolutionRatePercent": 20.833333333333336,
            "standardErrorPercent": 8.289816934939807,
            "totalCostUsd": 584.5591491999999,
            "totalTokens": 1224726937
          },
          "engineering": {
            "nTrials": 27,
            "passes": 4,
            "resolutionRatePercent": 14.814814814814813,
            "standardErrorPercent": 6.836727459191831,
            "totalCostUsd": 315.064889,
            "totalTokens": 569191613
          },
          "life": {
            "nTrials": 57,
            "passes": 10,
            "resolutionRatePercent": 17.543859649122805,
            "standardErrorPercent": 5.03775039171214,
            "totalCostUsd": 513.1874942,
            "totalTokens": 911217942
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 16,
            "resolutionRatePercent": 31.372549019607842,
            "standardErrorPercent": 6.497387212388346,
            "totalCostUsd": 1664.42933,
            "totalTokens": 3257195155
          },
          "physical": {
            "nTrials": 51,
            "passes": 12,
            "resolutionRatePercent": 23.52941176470588,
            "standardErrorPercent": 5.939743109381191,
            "totalCostUsd": 1144.6274403999996,
            "totalTokens": 2444538546
          }
        },
        "nTrials": 210,
        "passes": 47,
        "resolutionRatePercent": 22.380952380952383,
        "standardErrorPercent": 2.876164947101791,
        "totalCostUsd": 4221.8683028,
        "totalTokens": 8406870193
      },
      "model": {
        "display": {
          "label": "GPT-5.6 Sol",
          "url": "https://developers.openai.com/api/docs/models/gpt-5.6-sol"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "rank": 5,
      "reasoningEffort": "max",
      "releaseDate": "2026-07-09",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/378f8b49-dbd8-4acf-a065-8c07e1712de2"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "e3f2861a-97db-4730-a0ff-99f08d8c7299",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 6,
            "resolutionRatePercent": 25,
            "standardErrorPercent": 8.838834764831844,
            "totalCostUsd": 1622.574211,
            "totalTokens": 602637767
          },
          "engineering": {
            "nTrials": 27,
            "passes": 3,
            "resolutionRatePercent": 11.11111111111111,
            "standardErrorPercent": 6.04812282168686,
            "totalCostUsd": 2388.1075905000002,
            "totalTokens": 1317986164
          },
          "life": {
            "nTrials": 57,
            "passes": 10,
            "resolutionRatePercent": 17.543859649122805,
            "standardErrorPercent": 5.03775039171214,
            "totalCostUsd": 1963.0306967499992,
            "totalTokens": 864297832
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 17,
            "resolutionRatePercent": 33.33333333333333,
            "standardErrorPercent": 6.600983619844496,
            "totalCostUsd": 4627.1448445,
            "totalTokens": 2079346609
          },
          "physical": {
            "nTrials": 51,
            "passes": 9,
            "resolutionRatePercent": 17.647058823529413,
            "standardErrorPercent": 5.3381483116060595,
            "totalCostUsd": 3575.090842499999,
            "totalTokens": 1499293809
          }
        },
        "nTrials": 210,
        "passes": 45,
        "resolutionRatePercent": 21.428571428571427,
        "standardErrorPercent": 2.831517739900328,
        "totalCostUsd": 14175.948185250001,
        "totalTokens": 6363562181
      },
      "model": {
        "display": {
          "label": "Fable 5",
          "url": "https://www.anthropic.com/"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "rank": 6,
      "reasoningEffort": "max",
      "releaseDate": "2026-06-09",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/e3f2861a-97db-4730-a0ff-99f08d8c7299"
    },
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "58d4c48d-791b-4873-ba3e-af1cf997a79c",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 2,
            "resolutionRatePercent": 8.333333333333332,
            "standardErrorPercent": 5.64169333655275,
            "totalCostUsd": 39.16776078,
            "totalTokens": 1777639333
          },
          "engineering": {
            "nTrials": 27,
            "passes": 5,
            "resolutionRatePercent": 18.51851851851852,
            "standardErrorPercent": 7.4756798866604734,
            "totalCostUsd": 69.74610236399998,
            "totalTokens": 3041093185
          },
          "life": {
            "nTrials": 57,
            "passes": 6,
            "resolutionRatePercent": 10.526315789473683,
            "standardErrorPercent": 4.064886556427542,
            "totalCostUsd": 64.22627608800002,
            "totalTokens": 2470912103
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 6,
            "resolutionRatePercent": 11.76470588235294,
            "standardErrorPercent": 4.511558757925709,
            "totalCostUsd": 93.96861560400001,
            "totalTokens": 2921282541
          },
          "physical": {
            "nTrials": 51,
            "passes": 14,
            "resolutionRatePercent": 27.450980392156865,
            "standardErrorPercent": 6.2489869213841,
            "totalCostUsd": 118.824719418,
            "totalTokens": 4648478757
          }
        },
        "nTrials": 210,
        "passes": 33,
        "resolutionRatePercent": 15.714285714285714,
        "standardErrorPercent": 2.511392893650442,
        "totalCostUsd": 385.933474254,
        "totalTokens": 14859405919
      },
      "model": {
        "display": {
          "label": "DeepSeek V4.1 Flash",
          "url": "https://api-docs.deepseek.com/updates/"
        },
        "organization": {
          "label": "DeepSeek",
          "url": "https://www.deepseek.com/"
        },
        "version": null
      },
      "rank": 7,
      "reasoningEffort": "max",
      "releaseDate": "2026-09-10",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/58d4c48d-791b-4873-ba3e-af1cf997a79c"
    },
    {
      "harness": {
        "display": {
          "label": "Grok Build",
          "url": "https://x.ai/"
        },
        "organization": {
          "label": "xAI",
          "url": "https://x.ai"
        },
        "version": null
      },
      "id": "360317df-baea-4ead-8a11-09db70a27d15",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 6,
            "resolutionRatePercent": 25,
            "standardErrorPercent": 8.838834764831844,
            "totalCostUsd": 314.11143000000004,
            "totalTokens": 437274242
          },
          "engineering": {
            "nTrials": 27,
            "passes": 7,
            "resolutionRatePercent": 25.925925925925924,
            "standardErrorPercent": 8.433704334123124,
            "totalCostUsd": 343.575272,
            "totalTokens": 464565188
          },
          "life": {
            "nTrials": 57,
            "passes": 7,
            "resolutionRatePercent": 12.280701754385964,
            "standardErrorPercent": 4.347321296345194,
            "totalCostUsd": 377.22269599999987,
            "totalTokens": 516732172
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 2,
            "resolutionRatePercent": 3.9215686274509802,
            "standardErrorPercent": 2.718052078759498,
            "totalCostUsd": 1355.7633679999997,
            "totalTokens": 1874600150
          },
          "physical": {
            "nTrials": 51,
            "passes": 8,
            "resolutionRatePercent": 15.686274509803921,
            "standardErrorPercent": 5.09241697494762,
            "totalCostUsd": 680.1440580000001,
            "totalTokens": 913055022
          }
        },
        "nTrials": 210,
        "passes": 30,
        "resolutionRatePercent": 14.285714285714286,
        "standardErrorPercent": 2.414726442081476,
        "totalCostUsd": 3070.816824,
        "totalTokens": 4206226774
      },
      "model": {
        "display": {
          "label": "Grok 4.7",
          "url": "https://x.ai/news/grok-4-7"
        },
        "organization": {
          "label": "xAI",
          "url": "https://x.ai"
        },
        "version": null
      },
      "rank": 8,
      "reasoningEffort": "xhigh",
      "releaseDate": "2026-09-21",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-21T20:47:10.285461+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/360317df-baea-4ead-8a11-09db70a27d15"
    },
    {
      "harness": {
        "display": {
          "label": "mini-SWE-agent",
          "url": "https://mini-swe-agent.com/latest/"
        },
        "organization": {
          "label": "SWE-agent",
          "url": "https://github.com/SWE-agent"
        },
        "version": null
      },
      "id": "409f68a6-d6a4-43c9-adb8-63679dc17a45",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 3,
            "resolutionRatePercent": 12.5,
            "standardErrorPercent": 6.750771560841522,
            "totalCostUsd": 117.57938257500001,
            "totalTokens": 1055679015
          },
          "engineering": {
            "nTrials": 27,
            "passes": 3,
            "resolutionRatePercent": 11.11111111111111,
            "standardErrorPercent": 6.04812282168686,
            "totalCostUsd": 70.25994607499999,
            "totalTokens": 520230489
          },
          "life": {
            "nTrials": 57,
            "passes": 8,
            "resolutionRatePercent": 14.035087719298245,
            "standardErrorPercent": 4.600772407769725,
            "totalCostUsd": 129.486532875,
            "totalTokens": 947480980
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 510.9776022,
            "totalTokens": 5225589584
          },
          "physical": {
            "nTrials": 51,
            "passes": 12,
            "resolutionRatePercent": 23.52941176470588,
            "standardErrorPercent": 5.939743109381191,
            "totalCostUsd": 288.6729866250001,
            "totalTokens": 2750669844
          }
        },
        "nTrials": 210,
        "passes": 26,
        "resolutionRatePercent": 12.380952380952381,
        "standardErrorPercent": 2.2728283787427124,
        "totalCostUsd": 1116.97645035,
        "totalTokens": 10499649912
      },
      "model": {
        "display": {
          "label": "Gemini 3.8 Flash",
          "url": "https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash"
        },
        "organization": {
          "label": "Google DeepMind",
          "url": "https://deepmind.google/"
        },
        "version": null
      },
      "rank": 9,
      "reasoningEffort": "high",
      "releaseDate": "2026-09-02",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/409f68a6-d6a4-43c9-adb8-63679dc17a45"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "e8bf7739-3c94-46a3-95a5-fa4d4444034e",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 1,
            "resolutionRatePercent": 4.166666666666666,
            "standardErrorPercent": 4.078937543219003,
            "totalCostUsd": 694.2622874999998,
            "totalTokens": 753694113
          },
          "engineering": {
            "nTrials": 27,
            "passes": 2,
            "resolutionRatePercent": 7.4074074074074066,
            "standardErrorPercent": 5.040102351405716,
            "totalCostUsd": 831.492595,
            "totalTokens": 1065198528
          },
          "life": {
            "nTrials": 57,
            "passes": 4,
            "resolutionRatePercent": 7.017543859649122,
            "standardErrorPercent": 3.383417933803951,
            "totalCostUsd": 806.35009475,
            "totalTokens": 780549951
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 4,
            "resolutionRatePercent": 7.8431372549019605,
            "standardErrorPercent": 3.7646418038888574,
            "totalCostUsd": 1901.8542739999996,
            "totalTokens": 2166018647
          },
          "physical": {
            "nTrials": 51,
            "passes": 11,
            "resolutionRatePercent": 21.568627450980394,
            "standardErrorPercent": 5.7593182040981175,
            "totalCostUsd": 1603.7576537499995,
            "totalTokens": 2004770718
          }
        },
        "nTrials": 210,
        "passes": 22,
        "resolutionRatePercent": 10.476190476190476,
        "standardErrorPercent": 2.1133008267654967,
        "totalCostUsd": 5837.716905,
        "totalTokens": 6770231957
      },
      "model": {
        "display": {
          "label": "Opus 4.8",
          "url": "https://www.anthropic.com/"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "rank": 10,
      "reasoningEffort": "max",
      "releaseDate": "2026-05-28",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/e8bf7739-3c94-46a3-95a5-fa4d4444034e"
    },
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "c79d5ef6-b3c6-40da-84b5-946c11099426",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 2,
            "resolutionRatePercent": 8.333333333333332,
            "standardErrorPercent": 5.64169333655275,
            "totalCostUsd": 302.40976619999986,
            "totalTokens": 1197812821
          },
          "engineering": {
            "nTrials": 27,
            "passes": 2,
            "resolutionRatePercent": 7.4074074074074066,
            "standardErrorPercent": 5.040102351405716,
            "totalCostUsd": 141.00018310000002,
            "totalTokens": 464143146
          },
          "life": {
            "nTrials": 57,
            "passes": 4,
            "resolutionRatePercent": 7.017543859649122,
            "standardErrorPercent": 3.383417933803951,
            "totalCostUsd": 297.9608035,
            "totalTokens": 1031960432
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 5,
            "resolutionRatePercent": 9.803921568627452,
            "standardErrorPercent": 4.163980181949191,
            "totalCostUsd": 450.6649798000001,
            "totalTokens": 1639432828
          },
          "physical": {
            "nTrials": 51,
            "passes": 5,
            "resolutionRatePercent": 9.803921568627452,
            "standardErrorPercent": 4.163980181949191,
            "totalCostUsd": 791.8832043,
            "totalTokens": 3285452189
          }
        },
        "nTrials": 210,
        "passes": 18,
        "resolutionRatePercent": 8.571428571428571,
        "standardErrorPercent": 1.9317811536651808,
        "totalCostUsd": 1983.9189368999998,
        "totalTokens": 7618801416
      },
      "model": {
        "display": {
          "label": "GPT-5.6 Terra",
          "url": "https://developers.openai.com/api/docs/models/gpt-5.6-terra"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "rank": 11,
      "reasoningEffort": "max",
      "releaseDate": "2026-07-09",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/c79d5ef6-b3c6-40da-84b5-946c11099426"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "78f18248-3ee9-4eb4-b7af-19019f1fc64f",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 1,
            "resolutionRatePercent": 4.166666666666666,
            "standardErrorPercent": 4.078937543219003,
            "totalCostUsd": 1404.890923,
            "totalTokens": 1588775692
          },
          "engineering": {
            "nTrials": 27,
            "passes": 2,
            "resolutionRatePercent": 7.4074074074074066,
            "standardErrorPercent": 5.040102351405716,
            "totalCostUsd": 788.2680359999999,
            "totalTokens": 992085164
          },
          "life": {
            "nTrials": 57,
            "passes": 6,
            "resolutionRatePercent": 10.526315789473683,
            "standardErrorPercent": 4.064886556427542,
            "totalCostUsd": 1310.9276260000004,
            "totalTokens": 1506421392
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 3,
            "resolutionRatePercent": 5.88235294117647,
            "standardErrorPercent": 3.294776668301199,
            "totalCostUsd": 1571.9234520000002,
            "totalTokens": 2053648887
          },
          "physical": {
            "nTrials": 51,
            "passes": 5,
            "resolutionRatePercent": 9.803921568627452,
            "standardErrorPercent": 4.163980181949191,
            "totalCostUsd": 1726.556125,
            "totalTokens": 2345166065
          }
        },
        "nTrials": 210,
        "passes": 17,
        "resolutionRatePercent": 8.095238095238095,
        "standardErrorPercent": 1.8822364227102866,
        "totalCostUsd": 2733.16185912,
        "totalTokens": 8486097200
      },
      "model": {
        "display": {
          "label": "GLM 5.3",
          "url": "https://z.ai/"
        },
        "organization": {
          "label": "Z.AI",
          "url": "https://z.ai"
        },
        "version": null
      },
      "rank": 12,
      "reasoningEffort": "max",
      "releaseDate": "2026-08-14",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/78f18248-3ee9-4eb4-b7af-19019f1fc64f"
    },
    {
      "harness": {
        "display": {
          "label": "Claude Code",
          "url": "https://claude.com/product/claude-code"
        },
        "organization": {
          "label": "Anthropic",
          "url": "https://www.anthropic.com"
        },
        "version": null
      },
      "id": "66e64f64-abbd-40a6-a470-2151df25472e",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 3,
            "resolutionRatePercent": 12.5,
            "standardErrorPercent": 6.750771560841522,
            "totalCostUsd": 342.59728249999995,
            "totalTokens": 380051652
          },
          "engineering": {
            "nTrials": 27,
            "passes": 1,
            "resolutionRatePercent": 3.7037037037037033,
            "standardErrorPercent": 3.6344694923159864,
            "totalCostUsd": 354.7335025000001,
            "totalTokens": 536303864
          },
          "life": {
            "nTrials": 57,
            "passes": 4,
            "resolutionRatePercent": 7.017543859649122,
            "standardErrorPercent": 3.383417933803951,
            "totalCostUsd": 591.3448354999999,
            "totalTokens": 612759967
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 1,
            "resolutionRatePercent": 1.9607843137254901,
            "standardErrorPercent": 1.9414657705424987,
            "totalCostUsd": 714.7604920000001,
            "totalTokens": 863209487
          },
          "physical": {
            "nTrials": 51,
            "passes": 6,
            "resolutionRatePercent": 11.76470588235294,
            "standardErrorPercent": 4.511558757925709,
            "totalCostUsd": 536.2298835,
            "totalTokens": 799277999
          }
        },
        "nTrials": 210,
        "passes": 15,
        "resolutionRatePercent": 7.142857142857142,
        "standardErrorPercent": 1.7771905411718545,
        "totalCostUsd": 1523.7995976,
        "totalTokens": 3191602969
      },
      "model": {
        "display": {
          "label": "Kimi K3",
          "url": "https://www.moonshot.ai/"
        },
        "organization": {
          "label": "Moonshot AI",
          "url": "https://www.moonshot.ai"
        },
        "version": null
      },
      "rank": 13,
      "reasoningEffort": "max",
      "releaseDate": "2026-07-16",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/66e64f64-abbd-40a6-a470-2151df25472e"
    },
    {
      "harness": {
        "display": {
          "label": "Grok Build",
          "url": "https://x.ai/"
        },
        "organization": {
          "label": "xAI",
          "url": "https://x.ai"
        },
        "version": null
      },
      "id": "231bc6c3-8f2b-455e-8c87-5a08ec200b05",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 1,
            "resolutionRatePercent": 4.166666666666666,
            "standardErrorPercent": 4.078937543219003,
            "totalCostUsd": 217.32519000000005,
            "totalTokens": 244927564
          },
          "engineering": {
            "nTrials": 27,
            "passes": 4,
            "resolutionRatePercent": 14.814814814814813,
            "standardErrorPercent": 6.836727459191831,
            "totalCostUsd": 231.56777400000001,
            "totalTokens": 276719284
          },
          "life": {
            "nTrials": 57,
            "passes": 4,
            "resolutionRatePercent": 7.017543859649122,
            "standardErrorPercent": 3.383417933803951,
            "totalCostUsd": 281.65276,
            "totalTokens": 367523179
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 3,
            "resolutionRatePercent": 5.88235294117647,
            "standardErrorPercent": 3.294776668301199,
            "totalCostUsd": 2074.054437999999,
            "totalTokens": 2173354326
          },
          "physical": {
            "nTrials": 51,
            "passes": 3,
            "resolutionRatePercent": 5.88235294117647,
            "standardErrorPercent": 3.294776668301199,
            "totalCostUsd": 539.5871699999999,
            "totalTokens": 646860649
          }
        },
        "nTrials": 210,
        "passes": 15,
        "resolutionRatePercent": 7.142857142857142,
        "standardErrorPercent": 1.7771905411718545,
        "totalCostUsd": 3344.187332,
        "totalTokens": 3709385002
      },
      "model": {
        "display": {
          "label": "Grok 4.6",
          "url": "https://x.ai/"
        },
        "organization": {
          "label": "xAI",
          "url": "https://x.ai"
        },
        "version": null
      },
      "rank": 14,
      "reasoningEffort": "xhigh",
      "releaseDate": "2026-08-12",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/231bc6c3-8f2b-455e-8c87-5a08ec200b05"
    },
    {
      "harness": {
        "display": {
          "label": "mini-SWE-agent",
          "url": "https://mini-swe-agent.com/latest/"
        },
        "organization": {
          "label": "SWE-agent",
          "url": "https://github.com/SWE-agent"
        },
        "version": null
      },
      "id": "5a71d088-1bfd-4bbe-a65a-973a4fffd50f",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 63.50667502499999,
            "totalTokens": 503645605
          },
          "engineering": {
            "nTrials": 27,
            "passes": 2,
            "resolutionRatePercent": 7.4074074074074066,
            "standardErrorPercent": 5.040102351405716,
            "totalCostUsd": 56.408736075,
            "totalTokens": 377209430
          },
          "life": {
            "nTrials": 57,
            "passes": 5,
            "resolutionRatePercent": 8.771929824561402,
            "standardErrorPercent": 3.746919739866593,
            "totalCostUsd": 100.54085647500003,
            "totalTokens": 704804016
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 1,
            "resolutionRatePercent": 1.9607843137254901,
            "standardErrorPercent": 1.9414657705424987,
            "totalCostUsd": 585.3949539000001,
            "totalTokens": 5702475089
          },
          "physical": {
            "nTrials": 51,
            "passes": 4,
            "resolutionRatePercent": 7.8431372549019605,
            "standardErrorPercent": 3.7646418038888574,
            "totalCostUsd": 490.735906425,
            "totalTokens": 4734943451
          }
        },
        "nTrials": 210,
        "passes": 12,
        "resolutionRatePercent": 5.714285714285714,
        "standardErrorPercent": 1.6017483159468233,
        "totalCostUsd": 1296.5871279,
        "totalTokens": 12023077591
      },
      "model": {
        "display": {
          "label": "Gemini 3.7 Flash",
          "url": "https://ai.google.dev/gemini-api/docs/models/gemini-3.7-flash"
        },
        "organization": {
          "label": "Google DeepMind",
          "url": "https://deepmind.google/"
        },
        "version": null
      },
      "rank": 15,
      "reasoningEffort": "high",
      "releaseDate": "2026-08-13",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/5a71d088-1bfd-4bbe-a65a-973a4fffd50f"
    },
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "00b5eabf-caca-435c-b8a8-458633192aa8",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 71.822495184,
            "totalTokens": 1481972810
          },
          "engineering": {
            "nTrials": 27,
            "passes": 3,
            "resolutionRatePercent": 11.11111111111111,
            "standardErrorPercent": 6.04812282168686,
            "totalCostUsd": 108.39618107999999,
            "totalTokens": 2200281229
          },
          "life": {
            "nTrials": 57,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 107.43456482000002,
            "totalTokens": 2083306722
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 3,
            "resolutionRatePercent": 5.88235294117647,
            "standardErrorPercent": 3.294776668301199,
            "totalCostUsd": 146.126849924,
            "totalTokens": 2672070982
          },
          "physical": {
            "nTrials": 51,
            "passes": 2,
            "resolutionRatePercent": 3.9215686274509802,
            "standardErrorPercent": 2.718052078759498,
            "totalCostUsd": 140.452167108,
            "totalTokens": 2763809854
          }
        },
        "nTrials": 210,
        "passes": 8,
        "resolutionRatePercent": 3.8095238095238093,
        "standardErrorPercent": 1.3209662936780717,
        "totalCostUsd": 1120.15076932,
        "totalTokens": 11201441597
      },
      "model": {
        "display": {
          "label": "DeepSeek V4 Pro",
          "url": "https://api-docs.deepseek.com/news/news260813/"
        },
        "organization": {
          "label": "DeepSeek",
          "url": "https://www.deepseek.com/"
        },
        "version": null
      },
      "rank": 16,
      "reasoningEffort": "max",
      "releaseDate": "2026-08-13",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/00b5eabf-caca-435c-b8a8-458633192aa8"
    },
    {
      "harness": {
        "display": {
          "label": "Codex",
          "url": "https://openai.com/codex/"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "id": "7cf10b06-3e5b-4723-b98f-c679e773959b",
      "metrics": {
        "domains": {
          "earth": {
            "nTrials": 24,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 41.411879639999995,
            "totalTokens": 1640103008
          },
          "engineering": {
            "nTrials": 27,
            "passes": 2,
            "resolutionRatePercent": 7.4074074074074066,
            "standardErrorPercent": 5.040102351405716,
            "totalCostUsd": 35.81521945,
            "totalTokens": 1247416609
          },
          "life": {
            "nTrials": 57,
            "passes": 1,
            "resolutionRatePercent": 1.7543859649122806,
            "standardErrorPercent": 1.7389285185380774,
            "totalCostUsd": 83.50810593999996,
            "totalTokens": 2852715005
          },
          "mathematical": {
            "nTrials": 51,
            "passes": 4,
            "resolutionRatePercent": 7.8431372549019605,
            "standardErrorPercent": 3.7646418038888574,
            "totalCostUsd": 105.62130739999999,
            "totalTokens": 3739866520
          },
          "physical": {
            "nTrials": 51,
            "passes": 0,
            "resolutionRatePercent": 0,
            "standardErrorPercent": 0,
            "totalCostUsd": 116.84312716999993,
            "totalTokens": 4766821210
          }
        },
        "nTrials": 210,
        "passes": 7,
        "resolutionRatePercent": 3.3333333333333335,
        "standardErrorPercent": 1.2387055882620108,
        "totalCostUsd": 383.19963959999995,
        "totalTokens": 14246922352
      },
      "model": {
        "display": {
          "label": "GPT-5.6 Luna",
          "url": "https://developers.openai.com/api/docs/models/gpt-5.6-luna"
        },
        "organization": {
          "label": "OpenAI",
          "url": "https://openai.com"
        },
        "version": null
      },
      "rank": 17,
      "reasoningEffort": "max",
      "releaseDate": "2026-07-09",
      "safeguardMode": null,
      "sourceUpdatedAt": "2026-09-16T13:45:27.369938+00:00",
      "sourceUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0/leaderboards/v0-1-eval/rows/7cf10b06-3e5b-4723-b98f-c679e773959b"
    }
  ],
  "schemaVersion": 1,
  "source": {
    "announcementUrl": "https://www.terminal-bench-science.ai/announcement",
    "datasetVersionId": "2b817f26-dc4f-4477-8032-2218dcc553b5",
    "leaderboardApiUrl": "https://www.terminal-bench-science.ai/api/leaderboard?package=terminal-bench-science%2Fterminal-bench-science&name=v0-1-eval",
    "leaderboardId": "9a545a14-78d0-44b4-9003-afe062522a47",
    "leaderboardName": "v0-1-eval",
    "leaderboardPackage": "terminal-bench-science/terminal-bench-science",
    "leaderboardUpdatedAt": "2026-09-23T16:29:34.906395+00:00",
    "leaderboardUrl": "https://hub.harborframework.com/datasets/terminal-bench-science/terminal-bench-science/0.1.0?leaderboard=v0-1-eval&tab=leaderboard",
    "method": "version-pinned-release-owner-leaderboard-api",
    "name": "Terminal-Bench-Science and Harbor Framework",
    "releaseCommit": "f81afac4f11048e77a15dfc8fb1dbfb897fea0ce",
    "releaseCommittedAt": "2026-08-26T10:23:00Z",
    "releaseCommitUrl": "https://github.com/harbor-framework/terminal-bench-science/commit/f81afac4f11048e77a15dfc8fb1dbfb897fea0ce",
    "releaseDoiUrl": "https://doi.org/10.5281/zenodo.22110254",
    "releaseNotesUrl": "https://github.com/harbor-framework/terminal-bench-science/blob/f81afac4f11048e77a15dfc8fb1dbfb897fea0ce/release-notes-v0.1.0.md",
    "releasePublishedAt": "2026-08-26T10:24:33Z",
    "releaseTag": "v0.1.0",
    "releaseUrl": "https://github.com/harbor-framework/terminal-bench-science/releases/tag/v0.1.0",
    "repositoryUrl": "https://github.com/harbor-framework/terminal-bench-science",
    "retrievedAt": "2026-09-25T15:34:56.683Z",
    "sourceClass": "benchmark-owner"
  }
}
