{
  "generated": "2026-09-08",
  "benchmark": "drop",
  "results": [
    {
      "system": "NAQANet (paper baseline)",
      "developer": "AI2 / UC Irvine (Dua et al.)",
      "value": 47.0,
      "date": "2019-03-01",
      "source": {
        "url": "https://arxiv.org/abs/1903.00161",
        "kind": "paper",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "notes": "Complete model in Table 4, test F1 47.01; best prior baseline 32.7; expert humans 96.4."
      }
    },
    {
      "system": "o1",
      "developer": "OpenAI",
      "value": 90.2,
      "date": "2024-12-17",
      "source": {
        "url": "https://github.com/openai/simple-evals",
        "kind": "developer-report",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "shots": 3,
        "notes": "simple-evals README, DROP F1 3-shot; o3-high 89.8 (April 2025)."
      }
    },
    {
      "system": "DeepSeek-V4-Pro-Base",
      "developer": "DeepSeek",
      "value": 88.7,
      "date": "2026-04-22",
      "source": {
        "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro",
        "kind": "developer-report",
        "accessed": "2026-09-04"
      },
      "conditions": {
        "shots": 1,
        "notes": "DeepSeek-V4 model card, base-model table (HF repo created 2026-04-22). Base model, F1, 1-shot."
      }
    }
  ]
}