GPQA Diamond
General QA · 2023-11-20
The hardest GPQA subset, with graduate-level Google-proof questions in biology, physics, and chemistry that require deep domain reasoning. Scores are reported as accuracy.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 96.0 |
| GPT-5.5 | 95.5 |
| GPT-5.6 Sol | 94.9 |
| Opus 4.7 | 94.2 |
| Claude Mythos 5 | 94.1 |
| GPT-5.5 Pro | 93.9 |
| Kimi K3 | 93.4 |
| GPT-5.6 Terra | 93.3 |
| GPT-5.2 Pro | 93.2 |
| Grok 4.3 | 92.9 |
| Grok 4.5 | 92.9 |
| MiniMax M3 | 92.9 |
| Claude Mythos Preview | 92.9 |
| GPT-5.4 | 92.8 |
| GPT-5.3-Codex | 92.6 |
| Claude Fable 5 | 92.6 |
| Claude Opus 5 | 92.4 |
| Gemini 3.6 Flash | 92.4 |
| GPT-5.6 Luna | 92.4 |
| Interfaze Beta | 92.4 |
| Opus 4.8 | 92.0 |
| GPT-5.2 | 91.9 |
| Muse Spark 1.1 | 91.9 |
| Gemini 3.5 Flash | 91.4 |
| GLM-5.2 | 91.4 |
| Opus 4.6 | 91.3 |
| Gemini 3 Pro Preview | 90.9 |
| Sonnet 5 | 90.5 |
| DeepSeek-V4-Pro | 90.4 |
| Sonnet 4.6 | 89.9 |
| Grok 4.20 | 89.9 |
| Kimi K2.6 | 89.9 |
| Qwen3.7-Max | 89.9 |
| Kimi K2.7 Code | 89.5 |
| Inkling-Small | 89.5 |
| Gemini 3 Flash Preview | 89.4 |
| Qwen3.6-Max-Preview | 89.1 |
| Muse Spark | 88.4 |
| Qwen3.6 Plus (2026-04-02) | 88.4 |
| Inkling-Small Preview | 88.3 |