OpenCompass LLM - Agent (CompassBench 2601): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 (Thinking) | 76.2 |
| 2 | GLM-4.7 | 75.3 |
| 3 | Gemini 3 Pro (Preview) | 74.4 |
| 4 | Grok 4.1 | 73.1 |
| 5 | Claude Sonnet 4.5 (Thinking) | 72.6 |
| 6 | Kimi K2 (Thinking) | 71.4 |
| 7 | Qwen 3 Max (2026-01-23) | 70.9 |
| 8 | Tencent HY 2.0 Think | 68.9 |
| 9 | Kimi K2.5 | 68.8 |
| 10 | GPT-5.2 (High) | 68.1 |
| 11 | DeepSeek V3.2 | 67.9 |
| 12 | ERNIE 5.0 | 66.9 |
| 13 | MiniMax-M2.1 | 66.2 |
| 14 | O3 (2025-04-16) (High) | 60.6 |
| 15 | O4 Mini (2025-04-16) (High) | 57.7 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-agent-compassbench-2601 · How It Works · Data refreshed daily, snapshot 2026-09-19.