physics benchmark
GPQA Leaderboard
GPQA is a dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry.
Updated Sep 5, 2026
Models100
Model coverage247
MetricScore
EvidenceC
GPQA Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score96.00% | Percentile100.00% | Participants247 | EvidenceC | Evaluated |
| Rank02 | ModelAN | Score94.60% | Percentile99.59% | Participants247 | EvidenceC | Evaluated |
| Rank03 | ModelOP | Score94.60% | Percentile99.19% | Participants247 | EvidenceC | Evaluated |
| Rank04 | ModelGO | Score94.30% | Percentile98.78% | Participants247 | EvidenceC | Evaluated |
| Rank05 | ModelAN | Score94.20% | Percentile98.37% | Participants247 | EvidenceC | Evaluated |
| Rank06 | ModelAN | Score93.60% | Percentile97.97% | Participants247 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score93.60% | Percentile97.56% | Participants247 | EvidenceC | Evaluated |
| Rank08 | ModelMA | Score93.50% | Percentile97.15% | Participants247 | EvidenceC | Evaluated |
| Rank09 | ModelOP | Score93.20% | Percentile96.75% | Participants247 | EvidenceC | Evaluated |
| Rank10 | ModelXA | Score93.00% | Percentile96.34% | Participants247 | EvidenceB | Evaluated |
| Rank11 | ModelOP | Score92.90% | Percentile95.94% | Participants247 | EvidenceC | Evaluated |
| Rank12 | ModelOP | Score92.80% | Percentile95.53% | Participants247 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score92.60% | Percentile95.12% | Participants247 | EvidenceC | Evaluated |
| Rank14 | ModelOP | Score92.40% | Percentile94.72% | Participants247 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score92.40% | Percentile94.31% | Participants247 | EvidenceC | Evaluated |
| Rank16 | ModelOP | Score92.30% | Percentile93.90% | Participants247 | EvidenceC | Evaluated |
| Rank17 | ModelTE | Score92.30% | Percentile93.50% | Participants247 | EvidenceC | Evaluated |
| Rank18 | ModelGO | Score91.90% | Percentile93.09% | Participants247 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score91.70% | Percentile92.68% | Participants247 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score91.70% | Percentile92.28% | Participants247 | EvidenceC | Evaluated |
| Rank21 | ModelAN | Score91.30% | Percentile91.87% | Participants247 | EvidenceC | Evaluated |
| Rank22 | ModelZA | Score91.20% | Percentile91.46% | Participants247 | EvidenceC | Evaluated |
| Rank23 | ModelMA | Score90.50% | Percentile91.06% | Participants247 | EvidenceC | Evaluated |
| Rank24 | ModelGO | Score90.40% | Percentile90.65% | Participants247 | EvidenceC | Evaluated |
| Rank25 | ModelTE | Score90.40% | Percentile90.24% | Participants247 | EvidenceC | Evaluated |
| Rank26 | ModelAC | Score90.40% | Percentile89.84% | Participants247 | EvidenceC | Evaluated |
| Rank27 | ModelAC | Score90.30% | Percentile89.43% | Participants247 | EvidenceC | Evaluated |
| Rank28 | ModelDE | Score90.10% | Percentile89.02% | Participants247 | EvidenceC | Evaluated |
| Rank29 | ModelAN | Score89.90% | Percentile88.62% | Participants247 | EvidenceC | Evaluated |
| Rank30 | ModelTM | Score89.50% | Percentile88.21% | Participants247 | EvidenceC | Evaluated |
GPQA Highlights
The leading models and scores on this benchmark.
GPQA Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for GPQA
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is GPQA?
What GPQA measures and how its scores work.
GPQA is a benchmark dataset of 448 multiple-choice questions in biology, physics, and chemistry.
GPQA measures Score as a ratio based on performance on its questions.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of C.
- Family
- GPQA
- Modality
- text
- Primary category
- physics
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about GPQA.
Which model scores highest on GPQA?
GPT-6 Astra is currently ranked first with 96.00%.
What are the top three models on GPQA?
The current leaders are GPT-6 Astra (96.00%), Claude Mythos Preview (94.60%), and GPT-5.6 Sol (94.60%).
Which GPQA model has the lowest official input price?
DeepSeek-V4-Flash-Max has the lowest matched official input price at $0.14 input / $0.28 output per 1M tokens.
Which models are fastest among GPQA results?
The fastest matched records are GPT OSS 120B (500.00 tok/s via Groq), GPT-5.5 Instant (206.89 tok/s via OpenAI), and GPT-5 mini (167.61 tok/s via OpenAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.