OpenCompass LLM - Agent (CompassBench 2402): leaderboard
OpenCompass CompassBench 2402 edition (file dated 2024-02-20): agent (tool use) category 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. 35 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 Turbo (Preview) | 80.5 | #416 |
| 2 | GPT-3.5 Turbo | 73.5 | #858 |
| 3 | GLM-4 | 72.1 | #647 |
| 4 | Qwen Max | 66 | #388 |
| 5 | MiniMax-abab5.5 | 63.8 | #1030 |
| 6 | Qwen 1.5 14B Chat | 63.2 | #976 |
| 7 | Qwen-72B-Chat | 59.3 | #835 |
| 8 | internlm2-chat-20B | 58 | #988 |
| 9 | Qwen-14B-Chat | 57 | #1161 |
| 10 | internlm2-chat-7B | 54.7 | #1113 |
| 11 | Yi Chat 34B | 54.2 | #950 |
| 12 | Mixtral 8x7B Instruct | 53.8 | #1085 |
| 13 | Qwen 1.5 4B Chat | 52.6 | #1442 |
| 14 | Qwen-7B Chat | 50.2 | #1353 |
| 15 | Qwen 1.5 7B Chat | 48.4 | #1292 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-agent-compassbench-2402 · How It Works · Data refreshed daily, snapshot 2026-11-05.