OpenCompass Agent - Tool Use — leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Pro | 55.3 |
| 2 | GPT-5.4 (High) | 52.8 |
| 3 | DeepSeek V4 Flash | 46.7 |
| 4 | Claude Opus 4.7 (High) | 43.3 |
| 5 | Gemini 3.1 Pro (Preview) | 41 |
| 6 | GLM-5.1 | 39.9 |
| 7 | Kimi K2.6 | 39.3 |
| 8 | Qwen 3.6 Max Preview | 30.5 |
| 9 | Qwen 3.5 397B A17B | 28.7 |
| 10 | Hy3-preview (High) | 28.7 |
| 11 | Claude Sonnet 4.6 (High) | 24.7 |
| 12 | MiniMax-M2.7 | 23.7 |
| 13 | Qwen 3.6 27B | 23.3 |
| 14 | GPT-5.4 Mini (High) | 22 |
| 15 | Step 3.5 Flash | 18 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-agent-tool-use · How the rankings work · Data refreshed daily, snapshot 2026-07-22.