AA MMLU-Pro: leaderboard
Artificial Analysis independent evaluation of MMLU-Pro: 12,000 graduate-level questions across 14 subjects with 10 answer options.
Metric: Accuracy (%). Source: artificialanalysis.ai. Saturation forecast: Estimated already saturated. 349 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) (High) | 89.76 | #66 (Gemini 3 Pro (Preview)) |
| 2 | Gemini 3 Pro (Preview) (Low) | 89.54 | #66 (Gemini 3 Pro (Preview)) |
| 3 | Claude Opus 4.5 (Thinking) | 89.45 | #83 (Claude Opus 4.5) |
| 4 | Claude Opus 4.5 (Non-reasoning) | 88.92 | #83 (Claude Opus 4.5) |
| 5 | Gemini 3 Flash (Preview) (Non-reasoning) | 88.22 | #82 (Gemini 3 Flash (Preview)) |
| 6 | Claude Opus 4.1 (Thinking) | 87.99 | #140 (Claude Opus 4.1) |
| 7 | MiniMax-M2.1 | 87.49 | #289 |
| 8 | Claude Sonnet 4.5 (Thinking) | 87.45 | #142 (Claude Sonnet 4.5) |
| 9 | GPT-5.2 (xHigh) | 87.36 | #110 (GPT-5.2) |
| 10 | Claude Opus 4 (Thinking) | 87.32 | #163 (Claude Opus 4) |
| 11 | GPT-5 (High) | 87.07 | #98 (GPT-5) |
| 12 | GPT-5.1 (High) | 87.04 | #135 (GPT-5.1) |
| 13 | GPT-5 (Medium) | 86.74 | #98 (GPT-5) |
| 14 | Grok 4 | 86.58 | #181 |
| 15 | GPT-5 Codex (High) | 86.49 | #127 (GPT-5 Codex) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aa-mmlu-pro · How It Works · Data refreshed daily, snapshot 2026-10-20.