OpenCompass Code - HumanEval-X (CompassBench 2405): leaderboard
OpenCompass CompassBench 2405 edition (file dated 2024-06-12): multi-language code generation (HumanEval-X) sub-task of the code category. 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-4o (2024-05-13) | 69.5 | #337 |
| 2 | Claude 3 Opus | 68.7 | #565 |
| 3 | GPT-4 Turbo | 66.6 | #417 |
| 4 | DeepSeek V2 Chat | 60.1 | #572 |
| 5 | Mixtral 8x22B Instruct | 57 | #738 |
| 6 | GLM-4 | 54.5 | #647 |
| 7 | Yi 1.5 34B Chat | 54.1 | #798 |
| 8 | Yi-1.5-9B Chat | 53.9 | #1032 |
| 9 | Mistral Large | 53.9 | #713 |
| 10 | Yi Large | 53 | #566 |
| 11 | Moonshot v1 8K | 51.8 | #622 |
| 12 | Qwen 1.5 110B Chat | 50.2 | #693 |
| 13 | DBRX | 47.8 | #1041 |
| 14 | Qwen 1.5 32B Chat | 46.3 | #820 |
| 15 | Yi-1.5-6B-Chat | 46.2 | #1241 |
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-x-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-11-05.