OpenCompass Code - HumanEval - Chinese (CompassBench 2405): leaderboard
OpenCompass CompassBench 2405 edition (file dated 2024-06-12): 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. 39 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 Turbo | 76.2 | #417 |
| 2 | GPT-4o (2024-05-13) | 67.7 | #337 |
| 3 | DeepSeek V2 Chat | 66.5 | #572 |
| 4 | Claude 3 Opus | 65.8 | #565 |
| 5 | Moonshot v1 8K | 64.6 | #622 |
| 6 | Llama 3 70B Instruct | 62.2 | #640 |
| 7 | GPT-3.5 Turbo | 62.2 | #858 |
| 8 | Yi-1.5-9B Chat | 62.2 | #1032 |
| 9 | Mistral Large | 61.6 | #713 |
| 10 | Qwen 1.5 110B Chat | 61 | #693 |
| 11 | Yi Large | 61 | #566 |
| 12 | Mixtral 8x22B Instruct | 58.5 | #738 |
| 13 | GLM-4 | 57.9 | #647 |
| 14 | Yi 1.5 34B Chat | 57.3 | #798 |
| 15 | internlm2-chat-20B | 52.4 | #988 |
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-2405 · How It Works · Data refreshed daily, snapshot 2026-11-05.