OpenCompass LLM - Instruction Following - Chinese (CompassBench 2501): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 | 85.2 |
| 2 | Gemini 2.0 Flash (001) | 74.6 |
| 3 | Gemini 2.0 Pro (Preview 02-05) | 72.3 |
| 4 | Yi Lightning | 70.6 |
| 5 | GLM-4 Plus | 68.8 |
| 6 | GPT-4o (2024-11-20) | 65.2 |
| 7 | DeepSeek V3 | 61 |
| 8 | MiniMax-Text-01 | 59.2 |
| 9 | Qwen 2.5 Max | 57.1 |
| 10 | Doubao-1.5-pro-32k-250115 | 55.6 |
| 11 | GPT-4o Mini (2024-07-18) | 53.8 |
| 12 | Qwen 2.5 72B Instruct | 51.9 |
| 13 | Claude 3.5 Sonnet (20241022) | 50.8 |
| 14 | Yi 1.5 34B Chat | 50.8 |
| 15 | Gemma 2 27B (IT) | 50.4 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-instruction-following-chinese-compassbench-2501 · How It Works · Data refreshed daily, snapshot 2026-09-19.