OpenCompass Subjective - Knowledge - English (CompassBench 2405): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 39 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o (2024-05-13) | 70.5 |
| 2 | GPT-4 Turbo | 59.6 |
| 3 | Claude 3 Opus | 53.1 |
| 4 | DeepSeek V2 Chat | 52.1 |
| 5 | GLM-4 | 50.2 |
| 6 | Command-R+ | 49.6 |
| 7 | Mistral Large | 45.5 |
| 8 | Qwen 1.5 110B Chat | 45.3 |
| 9 | Moonshot v1 8K | 45.1 |
| 10 | Llama 3 70B Instruct | 43.2 |
| 11 | Yi Large | 41 |
| 12 | Mixtral 8x7B Instruct (v0.1) | 40.8 |
| 13 | Mixtral 8x22B Instruct (v0.1) | 38.5 |
| 14 | Mistral 7B Instruct (v0.3) | 32 |
| 15 | Yi 1.5 34B Chat | 30.3 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-subjective-knowledge-english-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-09-19.