physics benchmark
GPQA
A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.
Updated Aug 7, 2026
Published models100
Registry coverage233
MetricScore
EvidenceC
GPQA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
Score distribution
Top published rows on the benchmark's original scale.
GPQA highlights
The top published results on this benchmark's own scale.
What is GPQA?
Definition and scoring fields from the benchmark registry.
A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level C.
- Family
- GPQA
- Modality
- text
- Primary category
- physics
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- gpqa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about GPQA.
Which model scores highest on GPQA?
Claude Mythos Preview is currently ranked first with 94.6%.
What does GPQA measure?
A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
100 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is marked as eligible for the current LLMBoard capability methodology.