OpenCompass LLM - Agent (CompassBench 2405): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 39 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3 Opus | 81.1 |
| 2 | Mistral Large | 80.8 |
| 3 | Yi Large | 76.2 |
| 4 | GPT-4 Turbo | 75.7 |
| 5 | GPT-4o (2024-05-13) | 74.9 |
| 6 | Command-R+ | 73.4 |
| 7 | GLM-4 | 73.4 |
| 8 | GPT-3.5 Turbo | 72 |
| 9 | Qwen 1.5 32B Chat | 72 |
| 10 | Yi 1.5 34B Chat | 71.4 |
| 11 | Mixtral 8x22B Instruct (v0.1) | 70.9 |
| 12 | DeepSeek V2 Chat | 69.1 |
| 13 | Llama 3 70B Instruct | 65.6 |
| 14 | Yi-1.5-9B Chat | 64.2 |
| 15 | Qwen 1.5 14B Chat | 63 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-agent-compassbench-2405 · How It Works · Data refreshed daily, snapshot 2026-09-19.