Science
GPQA Diamond
Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.
Rankings
Higher is better#194.4%
GPT-5.4-Pro
294.3%
Gemini 3.1 Pro
394.2%
Claude Opus 4.7
493.6%
GPT-5.5
593.5%
Kimi K3
692.8%
GPT-5.4
792.4%
GPT-5.2
792.4%
Qwen3.7-Max
991.9%
Gemini 3.0 Pro
1091.3%
Claude Opus 4.6
1191.2%
GLM-5.2
1290.5%
Kimi K2.6
1390.4%
Gemini 3.0 Flash
1490.1%
DeepSeek-V4-Pro
1589.9%
Claude Sonnet 4.6
1689.5%
Muse Spark
1788.4%
Qwen3.5OW
1888.1%
GPT-5.1
1888.1%
DeepSeek-V4-Flash
2087.6%
Kimi K2.5
2187%
Claude Opus 4.5
2286.4%
Gemini 2.5 Pro
2386.2%
GLM-5.1
2486%
GLM-5
2486%
Qwen3.6OW
2685.7%
GLM-4.7
2783.4%
Claude Sonnet 4.5
2880.9%
Claude Opus 4.1
2979.6%
Claude Opus 4
3079.1%
GLM-4.5
3175.7%
o1
3275.4%
Claude Sonnet 4
3375.1%
Kimi K2
3375.1%
Kimi K2 (0905)
3573.3%
o1-preview
3673%
Claude Haiku 4.5
3768%
Claude 3.7 Sonnet
3860%
o1-mini
3959.4%
Claude 3.5 Sonnet
4050.4%
Claude 3 Opus
4149.9%
GPT-4o
4242.5%
GPT-4 Turbo
4341.6%
Claude 3.5 Haiku
4440.4%
Claude 3 Sonnet
4540.2%
GPT-4o mini
4633.3%
Claude 3 Haiku
GPQA Diamond — frequently asked questions
- What is GPQA Diamond?
- Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.
- Which AI model scores highest on GPQA Diamond?
- GPT-5.4-Pro by OpenAI holds the best GPQA Diamond result among tracked models, at 94.4% (released Mar 5 2026). Higher scores are better on this benchmark.
- What are the top 5 models on GPQA Diamond?
- 1. GPT-5.4-Pro (OpenAI) — 94.4%; 2. Gemini 3.1 Pro (Google) — 94.3%; 3. Claude Opus 4.7 (Anthropic) — 94.2%; 4. GPT-5.5 (OpenAI) — 93.6%; 5. Kimi K3 (Moonshot AI) — 93.5%.
- How many models have a published GPQA Diamond score?
- 46 tracked models have a published GPQA Diamond score. Scores are the figures reported by each lab at that model's release, so this page is a record of results over time rather than a re-run leaderboard.
- What is the best open model on GPQA Diamond?
- Qwen3.5 by Qwen is the highest-ranked model with downloadable weights on GPQA Diamond, scoring 88.4% at rank 17 overall.