OpenCompass Math - Competition - Chinese (CompassBench 2504): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 05-06) | 80 |
| 2 | Qwen 3 32B (Thinking) | 70 |
| 3 | O4 Mini (High) | 70 |
| 4 | O3 Mini (High) | 65 |
| 5 | DeepSeek R1 | 60 |
| 6 | Qwen 3 235B A22B (Thinking) | 60 |
| 7 | Doubao-1.5-thinking-pro-250415 | 60 |
| 8 | Grok 3 | 55 |
| 9 | DeepSeek V3 (0324) | 55 |
| 10 | GLM-4 32B (0414) | 55 |
| 11 | ERNIE-X1-Turbo-32K | 55 |
| 12 | GPT-4.1 Mini | 45 |
| 13 | GPT-4.1 (2025-04-14) | 40 |
| 14 | Qwen 3 235B A22B | 35 |
| 15 | Claude 3.7 Sonnet (Thinking) | 35 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-math-competition-chinese-compassbench-2504 · How It Works · Data refreshed daily, snapshot 2026-09-19.