OpenCompass Math - Competition — leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 72.1 |
| 2 | Kimi K2.6 | 72.1 |
| 3 | DeepSeek V4 Pro | 71.2 |
| 4 | Claude Opus 4.7 (High) | 70.5 |
| 5 | GPT-5.4 (High) | 69.9 |
| 6 | Claude Sonnet 4.6 (High) | 67.6 |
| 7 | Hy3-preview (High) | 67.6 |
| 8 | DeepSeek V4 Flash | 67 |
| 9 | Qwen 3.5 397B A17B | 65.7 |
| 10 | Qwen 3.6 Max Preview | 65.1 |
| 11 | Step 3.5 Flash | 63.1 |
| 12 | GPT-5.4 Mini (High) | 59.9 |
| 13 | GLM-5.1 | 59 |
| 14 | Gemma 4 31B (IT) | 58 |
| 15 | Qwen 3.6 27B | 57.7 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-math-competition · How the rankings work · Data refreshed daily, snapshot 2026-07-22.