Benchmarks · knowledge
MMLU-Pro
MMLU-Pro: broad knowledge multiple choice; accuracy.
86 models · 110 results · data as of 2026-11-06 · artificialanalysis.ai ↗
Leaderboard
Gemini 3 Pro Preview (High)
All results
110 rows| # | Model | Effort | MMLU-Pro |
|---|---|---|---|
| 1 | Gemini 3 Pro | high | 89.8% |
| 2 | Opus 4.5 | 89.5% | |
| 3 | Gemini 3 Flash | 89.0% | |
| 4 | Opus 4.5 | none | 88.9% |
| 5 | Opus 4.1 | 88.0% | |
| 6 | MiniMax M2.1 | 87.5% | |
| 7 | Sonnet 4.5 | 87.5% | |
| 8 | GPT-5.2 | xhigh | 87.4% |
| 9 | Opus 4 | 87.3% | |
| 10 | GPT-5 | high | 87.1% |
| 11 | GPT-5.1 | high | 87.0% |
| 12 | GPT-5 | medium | 86.7% |
| 13 | Grok 4 | 86.6% | |
| 14 | GPT-5 Codex | high | 86.5% |
| 15 | Gemini 2.5 Pro | 86.2% | |
| 16 | GPT-5.1 Codex | high | 86.0% |
| 17 | GPT-5 | low | 86.0% |
| 18 | GPT-5.2 | medium | 85.9% |
| 19 | GLM 4.7 | 85.6% | |
| 20 | o3 | 85.3% | |
| 21 | DeepSeek V3.1 | 85.1% | |
| 22 | DeepSeek V3.2 | 85.0% | |
| 23 | Grok 4 Fast | 85.0% | |
| 24 | Kimi K2 | 84.8% | |
| 25 | MiMo V2 Flash | 84.3% | |
| 26 | Qwen 3 235B-A22B | 84.3% | |
| 27 | Qwen 3 Max | none | 84.1% |
| 28 | o1 | 84.1% | |
| 29 | GPT-5 Mini | high | 83.7% |
| 30 | DeepSeek V3.2 | none | 83.7% |