OpenCompass Code - HumanEval - Chinese (CompassBench 2407): leaderboard
OpenCompass CompassBench 2407 edition (file dated 2024-08-05): code completion (HumanEval) sub-task of the code category, Chinese questions. CompassBench is OpenCompass's periodic evaluation of chat models on a private question set; each edition is a separate board. Score 0-100, higher is better.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Estimated already saturated. 36 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Large | 83.5 | #713 |
| 2 | Claude 3.5 Sonnet (20240620) | 81.1 | #308 |
| 3 | Qwen 2 72B Instruct | 79.3 | #579 |
| 4 | GPT-4o Mini (2024-07-18) | 78.7 | #568 |
| 5 | GPT-4o (2024-05-13) | 78 | #337 |
| 6 | DeepSeek V2 Chat (0628) | 76.8 | #409 |
| 7 | GLM-4 (0520) | 76.2 | #612 |
| 8 | Moonshot v1 8K | 74.4 | #622 |
| 9 | Llama 3.1 70B Instruct | 72.6 | #567 |
| 10 | Qwen 1.5 110B Chat | 72 | #693 |
| 11 | Gemma 2 27B (IT) | 70.7 | #652 |
| 12 | Qwen 2 7B Instruct | 69.5 | #982 |
| 13 | Yi Large | 69.5 | #566 |
| 14 | GLM-4 9B Chat | 68.9 | #911 |
| 15 | Mixtral 8x22B Instruct | 67.1 | #738 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-code-humaneval-chinese-compassbench-2407 · How It Works · Data refreshed daily, snapshot 2026-11-05.