OpenCompass LLM - Reasoning - English (CompassBench 2501): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 | 83.8 |
| 2 | O1 (2024-12-17) | 82.2 |
| 3 | Doubao-1.5-pro-32k-250115 | 81.3 |
| 4 | O3 Mini (2025-01-31) | 79.9 |
| 5 | O1 Mini (2024-09-12) | 76.3 |
| 6 | Gemini 2.0 Pro (Preview 02-05) | 73.4 |
| 7 | GLM-Zero-Preview | 73 |
| 8 | QwQ 32B-Preview | 72.3 |
| 9 | Gemini 2.0 Flash (001) | 67.9 |
| 10 | Qwen 2.5 Max | 67.1 |
| 11 | Qwen 2.5 72B Instruct | 66.6 |
| 12 | DeepSeek V3 | 65.5 |
| 13 | Claude 3.5 Sonnet (20241022) | 62.9 |
| 14 | GLM-4 Plus | 61.9 |
| 15 | GPT-4o (2024-11-20) | 59.4 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-reasoning-english-compassbench-2501 · How It Works · Data refreshed daily, snapshot 2026-09-19.