OpenCompass Subjective - Creation (CompassBench 2405): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 39 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o (2024-05-13) | 65.4 |
| 2 | GPT-4 Turbo | 51.3 |
| 3 | DeepSeek V2 Chat | 48.8 |
| 4 | Yi Large | 47.1 |
| 5 | Qwen 1.5 110B Chat | 46.6 |
| 6 | Claude 3 Opus | 46.5 |
| 7 | GLM-4 | 45.9 |
| 8 | Moonshot v1 8K | 44.2 |
| 9 | Yi 1.5 34B Chat | 41.7 |
| 10 | Command-R+ | 38.9 |
| 11 | internlm2-chat-20B | 37.9 |
| 12 | Yi-1.5-9B Chat | 33.4 |
| 13 | Mistral Large | 33 |
| 14 | internlm2-chat-7B | 31.5 |
| 15 | Llama 3 70B Instruct | 30.8 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-subjective-creation-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-09-19.