OpenCompass LLM - Math (CompassBench 2510): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Grok 4 | 68 |
| 2 | Qwen 3 235B A22B 2507 (Thinking) | 66.8 |
| 3 | MiniMax-M2 | 65.6 |
| 4 | GPT-5 (Thinking) | 65.6 |
| 5 | Seed-OSS-36B-Instruct | 64.5 |
| 6 | GPT-OSS-120B (High) | 64 |
| 7 | O3 (2025-04-16) (High) | 64 |
| 8 | Qwen 3 Next 80B A3B (Thinking) | 63.8 |
| 9 | DeepSeek R1 0528 | 63 |
| 10 | GLM-4.6 | 63 |
| 11 | Hunyuan-T1-20250822 | 60.2 |
| 12 | O4 Mini (2025-04-16) (High) | 59.9 |
| 13 | Claude Sonnet 4.5 (Thinking) | 59.8 |
| 14 | Nanbeige3.5-Pro (Thinking) | 59.5 |
| 15 | Claude Haiku 4.5 (Thinking) | 58 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-math-compassbench-2510 · How It Works · Data refreshed daily, snapshot 2026-09-19.