Benchmarks · knowledge
MMLU-Pro
MMLU-Pro: broad knowledge multiple choice; accuracy.
58 models · 72 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
Opus 4.5 none
All results
72 rows| # | Model | Effort | MMLU-Pro |
|---|---|---|---|
| 1 | Opus 4.5 | none | 88.9% |
| 2 | MiniMax M2.1 | 87.5% | |
| 3 | GPT-5.2 | xhigh | 87.4% |
| 4 | GPT-5 | high | 87.1% |
| 5 | GPT-5.1 | high | 87.0% |
| 6 | GPT-5 | medium | 86.7% |
| 7 | Grok 4 | 86.6% | |
| 8 | GPT-5 Codex | high | 86.5% |
| 9 | Gemini 2.5 Pro | 86.2% | |
| 10 | GPT-5.1 Codex | high | 86.0% |
| 11 | GPT-5 | low | 86.0% |
| 12 | GPT-5.2 | medium | 85.9% |
| 13 | GLM 4.7 | 85.6% | |
| 14 | o3 | 85.3% | |
| 15 | Qwen 3 Max | none | 84.1% |
| 16 | o1 | 84.1% | |
| 17 | GPT-5 Mini | high | 83.7% |
| 18 | DeepSeek V3.2 | none | 83.7% |
| 19 | GLM 4.5 | 83.5% | |
| 20 | DeepSeek V3.1 | none | 83.3% |
| 21 | o4 mini | high | 83.2% |
| 22 | GPT-5 Mini | medium | 82.8% |
| 23 | Kimi K2 | none | 82.4% |
| 24 | MiniMax M2 | 82.0% | |
| 25 | GPT-5.1 Codex Mini | high | 82.0% |
| 26 | GLM 4.5 Air | 81.5% | |
| 27 | GPT-5.2 | none | 81.4% |
| 28 | Gemini 2.5 Flash | none | 80.9% |
| 29 | Llama 4 Maverick | none | 80.9% |
| 30 | gpt-oss-120b | high | 80.8% |