OpenCompass Language - Creation - Chinese (CompassBench 2507): leaderboard
OpenCompass CompassBench 2507 edition (file dated 2025-08-12): creative writing sub-task of the language category, Chinese questions. CompassBench is OpenCompass's periodic evaluation of chat models on a private question set; each edition is a separate board. Score 0-100, higher is better.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Around February 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Doubao-Seed-1.6-thinking-250715 | 91.3 | #169 |
| 2 | GPT-5 | 88.7 | #105 |
| 3 | Qwen 3 235B A22B 2507 Instruct | 86.2 | #315 |
| 4 | Grok 4 | 86.1 | #189 |
| 5 | O4 Mini (2025-04-16) (High) | 81.7 | #183 (O4 Mini (2025-04-16)) |
| 6 | Gemma 3 27B (IT) | 77.9 | #528 |
| 7 | Hunyuan-T1-20250711 | 70.2 | #277 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 70 | #263 |
| 9 | ERNIE-X1-Turbo-32K | 70 | #288 |
| 10 | DeepSeek R1 0528 | 69.5 | #232 |
| 11 | Kimi K2 | 68.2 | #252 |
| 12 | GLM 4.5 Air | 67.6 | #403 |
| 13 | Hunyuan A13B-Instruct | 67.5 | #680 |
| 14 | MiniMax M1 80k | 66.6 | #437 |
| 15 | GLM-4.5 | 66.4 | #283 |
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-creation-chinese-compassbench-2507 · How It Works · Data refreshed daily, snapshot 2026-11-05.