OpenCompass: leaderboard
OpenCompass — Shanghai AI Lab's aggregate leaderboard averaging core academic benchmarks (English and Chinese) across reasoning, knowledge, language and code.
Metric: Average Score. Source: rank.opencompass.org.cn. Saturation forecast: Around January 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 81.3 | #66 |
| 2 | GLM-5 | 79 | #141 |
| 3 | GPT-5 (2025-08-07) (High) | 78.8 | |
| 4 | Qwen 3.5 397B A17B | 78.4 | #146 |
| 5 | Kimi K2.5 | 78.2 | #144 |
| 6 | DeepSeek V3.2 Speciale | 78.2 | #128 |
| 7 | GLM-4.7 | 77.6 | #192 |
| 8 | Step 3.5 Flash | 76.9 | #270 |
| 9 | DeepSeek V3.2 | 75.3 | #205 |
| 10 | Kimi K2 (Thinking) | 75.3 | #243 (Kimi K2) |
| 11 | Gemini 2.5 Pro | 73.6 | #152 |
| 12 | MiniMax-M2.5 | 73.3 | #303 |
| 13 | GLM-4.6 | 73.3 | #252 |
| 14 | MiMo-V2-Flash | 73.3 | #372 |
| 15 | GPT-OSS-120B (High) | 73.2 | #341 (GPT-OSS-120B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass · How It Works · Data refreshed daily, snapshot 2026-10-20.