MATH500 math500 Leaderboard
Auto-discovered from papers reporting MATH500 (Accuracy). Β· Metric: Accuracy (higher is better) Β· π’ Updated 8h ago
| # | Model | Accuracy | Paper |
|---|---|---|---|
| 1 | Qwen/QwQ-32B | 90.60 | link |
| 2 | OpenThinker3-7B | 90.00 | link |
| 3 | deepseek-reasoner | 89.60 | link |
| 4 | Qwen-2.5-7B-Instruct | 89.00 | link |
| 5 | OpenThinker-7B | 84.40 | link |
| 6 | microsoft/Phi-4-reasoning-plus | 84.00 | link |
| 7 | Llama-3.1-8B-Instruct | 83.80 | link |
| 8 | w/ R1 template | 83.80 | link |
| 9 | FD + AD | 40.60 | link |
| 10 | FD | 40.40 | link |
| 11 | EB | 40.00 | link |
| 12 | EB + AD | 40.00 | link |
| 13 | Top-k | 28.40 | link |
| 14 | Top-k + AD | 28.40 | link |