OpenCompass Agent - Multi-Turn — leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 (Thinking) | 84.8 |
| 2 | Gemini 3 Pro (Preview) | 82.5 |
| 3 | GLM-4.7 | 81.8 |
| 4 | Claude Sonnet 4.5 (Thinking) | 81.2 |
| 5 | Grok 4.1 | 80.5 |
| 6 | DeepSeek V3.2 | 78.3 |
| 7 | Kimi K2 (Thinking) | 77 |
| 8 | ERNIE 5.0 | 75.2 |
| 9 | Qwen 3 Max (2026-01-23) | 75.2 |
| 10 | GPT-5.2 (High) | 72.5 |
| 11 | Kimi K2.5 | 72 |
| 12 | MiniMax-M2.1 | 70.5 |
| 13 | O3 (2025-04-16) (High) | 66.2 |
| 14 | O4 Mini (2025-04-16) (High) | 60.8 |
| 15 | GPT-OSS-120B (High) | 57.4 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-agent-multi-turn · How the rankings work · Data refreshed daily, snapshot 2026-07-22.