OpenCompass Research - MMLU-Pro — leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 48 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 89.3 |
| 2 | Qwen 3.5 397B A17B | 87.6 |
| 3 | Kimi K2.5 | 86.2 |
| 4 | GPT-5 (High) | 86.2 |
| 5 | Gemini 2.5 Pro | 85.8 |
| 6 | DeepSeek V3.2 | 85.8 |
| 7 | DeepSeek V3.2 Speciale | 85.5 |
| 8 | GLM-5 | 85.2 |
| 9 | Kimi K2 (Thinking) | 84.3 |
| 10 | GLM-4.7 | 84 |
| 11 | DeepSeek R1 0528 | 83.5 |
| 12 | Qwen 3 235B A22B (Thinking) | 83.5 |
| 13 | Step 3.5 Flash | 83.5 |
| 14 | MiMo-V2-Flash | 83.1 |
| 15 | doubao-seed-1.6-thinking-250615 | 83.1 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-research-mmlu-pro · How the rankings work · Data refreshed daily, snapshot 2026-07-22.