OpenCompass Language - Chat (CompassBench 2604): leaderboard
OpenCompass CompassBench 2604 edition (a fresh private question set per edition), chat (dialogue) subset of the language category. Score 0-100; higher is better. Pinned to the 26-04 edition because the newest CompassBench edition no longer publishes this table.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Estimated already saturated. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Doubao-Seed-2-0-Pro-260215 (High) | 91.7 | |
| 2 | Qwen 3.6 Max Preview | 87.1 | #91 |
| 3 | Kimi K2.6 | 86.7 | #113 |
| 4 | GPT-5.4 (2026-03-05) (High) | 84.6 | #63 (GPT-5.4 (2026-03-05)) |
| 5 | DeepSeek V4 Pro | 83.3 | #112 |
| 6 | Hy3-preview (High) | 82.9 | #166 (Hy3-preview) |
| 7 | Claude Opus 4.7 (High) | 82.5 | #51 (Claude Opus 4.7) |
| 8 | Claude Sonnet 4.6 (High) | 82.1 | #94 (Claude Sonnet 4.6) |
| 9 | MiniMax-M2.7 | 81.2 | #270 |
| 10 | GLM-5.1 | 80.8 | #109 |
| 11 | Qwen 3.5 397B A17B | 79.6 | #154 |
| 12 | Gemini 3.1 Pro (Preview) | 78.3 | #64 |
| 13 | Qwen 3.6 27B | 77.9 | #182 |
| 14 | Step 3.5 Flash | 77.5 | #281 |
| 15 | JT-MINI | 77.5 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-language-chat-compassbench-2604 · How It Works · Data refreshed daily, snapshot 2026-11-05.