OpenCompass LLM - Reasoning (CompassBench 2510): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.5 (Thinking) | 69.3 |
| 2 | GPT-5 (Thinking) | 69.3 |
| 3 | Gemini 2.5 Pro | 68.5 |
| 4 | O3 (2025-04-16) (High) | 68.4 |
| 5 | GLM-4.6 | 67.5 |
| 6 | Grok 4 | 65.5 |
| 7 | O4 Mini (2025-04-16) (High) | 65.5 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 65.3 |
| 9 | Qwen 3 Next 80B A3B (Thinking) | 64.7 |
| 10 | Seed-OSS-36B-Instruct | 64.5 |
| 11 | Hunyuan-T1-20250822 | 64.5 |
| 12 | DeepSeek V3.2 Exp | 63.8 |
| 13 | GPT-OSS-120B (High) | 63.7 |
| 14 | Ling-1T | 63.2 |
| 15 | DeepSeek R1 0528 | 62.5 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-reasoning-compassbench-2510 · How It Works · Data refreshed daily, snapshot 2026-09-19.