OpenCompass Subjective - Average (CompassBench 2405): leaderboard
OpenCompass CompassBench 2405 edition (file dated 2024-06-12): subjective open-ended evaluation, average score. 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: Estimated already saturated. 39 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-05-13) | 65.9 | #337 |
| 2 | GPT-4 Turbo | 53.9 | #417 |
| 3 | Claude 3 Opus | 49.1 | #565 |
| 4 | GLM-4 | 48.8 | #647 |
| 5 | Qwen 1.5 110B Chat | 47.9 | #693 |
| 6 | DeepSeek V2 Chat | 47.9 | #572 |
| 7 | Yi Large | 45.2 | #566 |
| 8 | Moonshot v1 8K | 43 | #622 |
| 9 | Nanbeige2-16B-Chat | 40.9 | #971 |
| 10 | Yi 1.5 34B Chat | 38.3 | #798 |
| 11 | Qwen 1.5 72B Chat | 38 | #782 |
| 12 | Mistral Large | 37.8 | #713 |
| 13 | Command-R+ | 37.2 | #869 |
| 14 | Llama 3 70B Instruct | 35.9 | #640 |
| 15 | Qwen 1.5 32B Chat | 33.9 | #820 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-subjective-average-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-11-05.