OpenCompass Language - Chat — leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Max Preview | 87.1 |
| 2 | Kimi K2.6 | 86.7 |
| 3 | GPT-5.4 (High) | 84.6 |
| 4 | DeepSeek V4 Pro | 83.3 |
| 5 | Hy3-preview (High) | 82.9 |
| 6 | Claude Opus 4.7 (High) | 82.5 |
| 7 | Claude Sonnet 4.6 (High) | 82.1 |
| 8 | MiniMax-M2.7 | 81.2 |
| 9 | GLM-5.1 | 80.8 |
| 10 | Qwen 3.5 397B A17B | 79.6 |
| 11 | Gemini 3.1 Pro (Preview) | 78.3 |
| 12 | Qwen 3.6 27B | 77.9 |
| 13 | Step 3.5 Flash | 77.5 |
| 14 | JT-MINI | 77.5 |
| 15 | DeepSeek V4 Flash | 77.1 |
Interactive version: aibenchmarks.dev/benchmark?slug=opencompass-language-chat · How the rankings work · Data refreshed daily, snapshot 2026-07-12.