SUPERChem
Science
Peking University's SUPERChem: 500 expert-curated, reasoning-intensive chemistry problems available in multimodal and text-only forms. Scores are final-answer accuracy.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 80.3 |
| Claude Opus 5.5 | 79.6 |
| GPT-6.1 Sol | 77.8 |
| GPT-5.6 Sol Pro | 74.9 |
| Claude Fable 5.1 | 74.7 |
| Claude Opus 5 | 74.7 |
| Gemini 3.7 Flash | 72.8 |
| GPT-6 Sol | 72.8 |
| Gemini 3.8 Flash | 72.6 |
| GPT-5.6 Sol | 71.0 |
| Gemini 3.5 Flash | 68.3 |
| GPT-5.5 | 66.8 |
| Gemini 3.6 Flash | 66.0 |
| Grok 4.6 | 65.7 |
| Grok 4.5 | 65.5 |
| GPT-5.6 Terra | 64.9 |
| Claude Fable 5 | 64.4 |
| GPT-5.4 | 64.1 |
| Kimi K3 | 63.7 |
| Claude Haiku 5.5 | 62.0 |
| Opus 4.8 | 60.3 |
| GPT-5.6 Luna | 58.1 |
| DeepSeek V4.1 Flash | 56.7 |
| GPT-6 Luna | 55.5 |
| Sonnet 5 | 55.3 |
| Kimi K2.7 Code | 51.5 |
| Opus 4.7 | 46.9 |
| Sonnet 4.6 | 46.3 |
| MiniMax M3 | 41.8 |