OpenCompass Language - Creation - English (CompassBench 2504): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 05-06) | 74.2 |
| 2 | Doubao-1.5-thinking-pro-250415 | 73.7 |
| 3 | Qwen 3 235B A22B (Thinking) | 70.4 |
| 4 | ERNIE-X1-Turbo-32K | 69.3 |
| 5 | O4 Mini (High) | 68.4 |
| 6 | Grok 3 | 67.7 |
| 7 | Qwen 3 235B A22B | 67.5 |
| 8 | Claude 3.7 Sonnet (Thinking) | 66.9 |
| 9 | DeepSeek R1 | 66.7 |
| 10 | Qwen 3 32B (Thinking) | 65.5 |
| 11 | Gemma 3 27B (IT) | 64.7 |
| 12 | O3 Mini (High) | 63 |
| 13 | GPT-4.1 (2025-04-14) | 61.5 |
| 14 | DeepSeek V3 (0324) | 60.9 |
| 15 | GPT-4.1 Mini | 60 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-language-creation-english-compassbench-2504 · How It Works · Data refreshed daily, snapshot 2026-09-19.