OpenCompass LLM - Agent (CompassBench 2510): leaderboard
OpenCompass CompassBench 2510 edition (file dated 2025-11-10): 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: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 (Thinking) | 67.2 | #148 (Claude Sonnet 4.5) |
| 2 | GLM-4.6 | 66.7 | #265 |
| 3 | Claude Haiku 4.5 (Thinking) | 63.4 | #287 (Claude Haiku 4.5) |
| 4 | Kimi K2 0905 | 62.5 | #292 |
| 5 | DeepSeek R1 0528 | 62.3 | #232 |
| 6 | O4 Mini (2025-04-16) (High) | 59.4 | #183 (O4 Mini (2025-04-16)) |
| 7 | Qwen 3 Next 80B A3B (Thinking) | 59.2 | #312 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 58.4 | #263 |
| 9 | Doubao-Seed-1.6-251015 (Thinking) | 58 | #208 (Doubao-Seed-1.6-251015) |
| 10 | O3 (2025-04-16) (High) | 57.4 | #125 (O3 (2025-04-16)) |
| 11 | Gemini 2.5 Pro | 57.3 | #159 |
| 12 | DeepSeek V3.2 Exp | 56.8 | #241 |
| 13 | MiniMax-M2 | 56.2 | #327 |
| 14 | Hunyuan-T1-20250822 | 52.4 | #306 |
| 15 | Seed-OSS-36B-Instruct | 50.5 | #384 |
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-2510 · How It Works · Data refreshed daily, snapshot 2026-11-05.