OpenCompass Code - Code QA - Chinese (CompassBench 2504): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 05-06) | 93.6 |
| 2 | Grok 3 | 90.2 |
| 3 | DeepSeek V3 (0324) | 89.7 |
| 4 | Doubao-1.5-thinking-pro-250415 | 88.7 |
| 5 | ERNIE-X1-Turbo-32K | 87.3 |
| 6 | Qwen 3 32B (Thinking) | 86.5 |
| 7 | MiniMax-Text-01 | 86.5 |
| 8 | DeepSeek R1 | 84.6 |
| 9 | Qwen 3 235B A22B | 84.1 |
| 10 | O4 Mini (High) | 83.8 |
| 11 | Claude 3.7 Sonnet (Thinking) | 83.1 |
| 12 | Qwen 3 235B A22B (Thinking) | 79.9 |
| 13 | Gemma 3 27B (IT) | 77.7 |
| 14 | GPT-4.1 (2025-04-14) | 75.5 |
| 15 | Step 2 16K | 75.5 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-code-code-qa-chinese-compassbench-2504 · How It Works · Data refreshed daily, snapshot 2026-09-19.