OpenCompass Language - Creation - Chinese (CompassBench 2501): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 | 82.4 |
| 2 | Qwen 2.5 Max | 76.5 |
| 3 | Gemini 2.0 Pro (Preview 02-05) | 75 |
| 4 | GPT-4o (2024-11-20) | 73.9 |
| 5 | O1 (2024-12-17) | 72.7 |
| 6 | O1 Mini (2024-09-12) | 70.5 |
| 7 | Gemini 2.0 Flash (001) | 70.3 |
| 8 | MiniMax-Text-01 | 69.7 |
| 9 | Yi Lightning | 69.3 |
| 10 | DeepSeek V3 | 69.2 |
| 11 | O3 Mini (2025-01-31) | 69 |
| 12 | InternLM3-8B-Instruct | 62.7 |
| 13 | Claude 3.5 Sonnet (20241022) | 62.5 |
| 14 | GLM-4 Plus | 61.7 |
| 15 | Doubao-1.5-pro-32k-250115 | 61.5 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-language-creation-chinese-compassbench-2501 · How It Works · Data refreshed daily, snapshot 2026-09-19.