OpenCompass Reasoning - Common Sense - Chinese (CompassBench 2411): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 34 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Yi Lightning | 71.2 |
| 2 | GLM-4 Plus | 69.2 |
| 3 | O1 Mini (2024-09-12) | 69 |
| 4 | GPT-4o (2024-11-20) | 68 |
| 5 | TeleChat2 | 67 |
| 6 | DeepSeek V2.5 | 64.4 |
| 7 | Grok Beta | 63.1 |
| 8 | Claude 3.5 Sonnet (20241022) | 62.6 |
| 9 | Qwen 2.5 72B Instruct | 61.4 |
| 10 | 360gpt2-pro | 58.5 |
| 11 | Mistral Large 2 (Nov) Instruct (2411) | 57.8 |
| 12 | Step 2 16K | 56.8 |
| 13 | GPT-4o (2024-08-06) | 50 |
| 14 | Qwen 2.5 7B Instruct | 49.8 |
| 15 | Gemma 2 27B (IT) | 45.3 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-reasoning-common-sense-chinese-compassbench-2411 · How It Works · Data refreshed daily, snapshot 2026-09-19.