Math
MATH-500
Competition math problems (500-problem set).
11 models published a score
| # | Model | Company | Score |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | Anthropic | 97.8 |
| 2 | DeepSeek R1 0528 | DeepSeek | 97.3 |
| 3 | Llama 4 Behemoth | Meta | 95.0 |
| 4 | Mistral Large 3 | Mistral AI | 93.6 |
| 5 | Reka Flash 3 | Reka | 89.3 |
| 6 | Llama 4 Maverick | Meta | 87.3 |
| 7 | Nemotron 3 Nano | Nvidia | 82.9 |
| 8 | Command A | Cohere | 80.0 |
| 9 | Nova Pro | Amazon | 76.6 |
| 10 | Yi-Lightning | 01.AI | 76.4 |
| 11 | Nova Lite | Amazon | 73.3 |