OpenCompass Code - Code QA - Chinese (CompassBench 2409): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 30 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V2.5 | 64.7 |
| 2 | Qwen 2.5 72B Instruct | 62.1 |
| 3 | GLM-4 Plus | 61.9 |
| 4 | Step 2 16K | 61.9 |
| 5 | GPT-4o (2024-05-13) | 58.2 |
| 6 | Mistral Large 2 (Jul) | 57 |
| 7 | Claude 3.5 Sonnet (20240620) | 56.2 |
| 8 | GPT-4o Mini (2024-07-18) | 55.7 |
| 9 | Gemini 1.5 Pro | 55.4 |
| 10 | GPT-4o (2024-08-06) | 54.1 |
| 11 | Qwen 2.5 7B Instruct | 50.8 |
| 12 | Gemma 2 27B (IT) | 49.7 |
| 13 | Llama 3.1 70B Instruct | 49.2 |
| 14 | Mistral-Small-Instruct-2409 | 47.6 |
| 15 | Mistral Nemo Instruct (2407) | 46.1 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-code-code-qa-chinese-compassbench-2409 · How It Works · Data refreshed daily, snapshot 2026-09-19.