{
  "generated": "2026-09-08",
  "benchmark": "gpqa-diamond",
  "results": [
    {
      "system": "GPT-4",
      "developer": "OpenAI",
      "value": 39.4,
      "date": "2023-11-20",
      "source": {
        "url": "https://arxiv.org/abs/2311.12022",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "shots": 0,
        "notes": "Table 1 of the GPQA paper."
      }
    },
    {
      "system": "o1-2024-12-17 (high)",
      "developer": "OpenAI",
      "value": 76.8,
      "date": "2025-02-13",
      "source": {
        "url": "https://epoch.ai/benchmarks/gpqa-diamond",
        "kind": "independent-evaluation",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Epoch AI Benchmarking Hub run (data export); gpt-4o-2024-08-06 scored 49.2 and claude-3-5-sonnet-20241022 55.3 in January 2025 runs."
      }
    },
    {
      "system": "o3-2025-04-16 (high)",
      "developer": "OpenAI",
      "value": 81.8,
      "date": "2025-04-16",
      "source": {
        "url": "https://epoch.ai/benchmarks/gpqa-diamond",
        "kind": "independent-evaluation",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Epoch AI Benchmarking Hub; OpenAI's own o3 launch figure was 83.3."
      }
    },
    {
      "system": "Gemini 3 Pro",
      "developer": "Google DeepMind",
      "value": 91.9,
      "date": "2025-11-18",
      "source": {
        "url": "https://blog.google/products/gemini/gemini-3/",
        "kind": "developer-report",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "tools": false,
        "notes": "Gemini 3 launch post; Epoch's independent run of gemini-3-pro-preview scored 92.6 the next day."
      }
    },
    {
      "system": "Gemini 3.1 Pro",
      "developer": "Google DeepMind",
      "value": 94.3,
      "date": "2026-05-01",
      "source": {
        "url": "https://benchlm.ai/benchmarks/gpqa-diamond",
        "kind": "aggregator",
        "accessed": "2026-09-04"
      }
    },
    {
      "system": "GPT-6 Astra",
      "developer": "OpenAI",
      "value": 96.0,
      "date": "2026-09-03",
      "source": {
        "url": "https://benchlm.ai/benchmarks/gpqa-diamond",
        "kind": "aggregator",
        "accessed": "2026-09-04"
      }
    },
    {
      "system": "GPT-6 Astra (max)",
      "developer": "OpenAI",
      "value": 95.8,
      "date": "2026-09-03",
      "source": {
        "url": "https://epoch.ai/benchmarks/gpqa-diamond",
        "kind": "independent-evaluation",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Epoch AI Benchmarking Hub run started 2026-08-30 (stderr 1.3); OpenAI's launch table reports 96.0, which benchlm.ai republishes."
      }
    }
  ]
}