OpenCompass LLM - Reasoning (CompassBench 2405): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 39 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 Turbo | 52.1 |
| 2 | GPT-4o (2024-05-13) | 47.8 |
| 3 | Yi Large | 46.8 |
| 4 | GLM-4 | 44 |
| 5 | Claude 3 Opus | 42.9 |
| 6 | DeepSeek V2 Chat | 41.5 |
| 7 | Qwen 1.5 32B Chat | 41.3 |
| 8 | Yi 1.5 34B Chat | 40.4 |
| 9 | Qwen 1.5 72B Chat | 40.3 |
| 10 | Llama 3 70B Instruct | 39.7 |
| 11 | Yi-1.5-9B Chat | 39.7 |
| 12 | Moonshot v1 8K | 39.1 |
| 13 | Qwen 1.5 14B Chat | 38.6 |
| 14 | Qwen 1.5 110B Chat | 37.5 |
| 15 | Mistral Large | 37.4 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-reasoning-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-09-19.