OpenCompass Language - Dialogue - Chinese (CompassBench 2510): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 96.1 |
| 2 | O3 (2025-04-16) (High) | 95.9 |
| 3 | GPT-5 (Thinking) | 95.9 |
| 4 | Qwen 3 235B A22B 2507 (Thinking) | 95.5 |
| 5 | GLM-4.6 | 94.6 |
| 6 | Hunyuan-T1-20250822 | 94.6 |
| 7 | O4 Mini (2025-04-16) (High) | 94.3 |
| 8 | Grok 4 | 93.7 |
| 9 | GPT-OSS-120B (High) | 93.3 |
| 10 | Seed-OSS-36B-Instruct | 93.2 |
| 11 | Nanbeige3.5-Pro (Thinking) | 92.8 |
| 12 | Kimi K2 0905 | 91.7 |
| 13 | DeepSeek R1 0528 | 91.4 |
| 14 | Ling-1T | 90.1 |
| 15 | Gemma 3 27B (IT) | 89.2 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-language-dialogue-chinese-compassbench-2510 · How It Works · Data refreshed daily, snapshot 2026-09-19.