{
  "id": "gpqa",
  "code": "PTL-0189",
  "term": "GPQA",
  "aliases": [
    "GPQA Diamond",
    "Graduate-Level Google-Proof Q&A"
  ],
  "category": "benchmarks",
  "definition": "GPQA (Graduate-Level Google-Proof Q&A) is a benchmark of difficult multiple-choice questions in biology, physics, and chemistry, written by domain experts so that skilled non-experts cannot answer them reliably even with unrestricted web access.",
  "description": "Its high-quality \"Diamond\" subset is the version most often reported in model announcements. Because the questions resist search, it measures expert-level reasoning more than retrieval. It is also used to study scalable oversight, the question of how people can supervise models that know more than they do.",
  "example": null,
  "broader": [
    "benchmark"
  ],
  "narrower": [],
  "related": [
    "mmlu-pro",
    "humanitys-last-exam",
    "reasoning-model"
  ],
  "introduced": 2023,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "GPQA: A Graduate-Level Google-Proof Q&A Benchmark",
      "authors": "Rein et al.",
      "year": 2023,
      "url": "https://arxiv.org/abs/2311.12022"
    }
  ],
  "url": "https://protologue.com/t/gpqa/",
  "citation": "Protologue. (2026). GPQA. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0189). https://protologue.com/t/gpqa/"
}