Vellum - MMMLU: leaderboard
Vellum's independently-run Multilingual MMLU evaluation. Knowledge questions across multiple languages.
Metric: Accuracy (%). Source: www.vellum.ai. Saturation forecast: Estimated already saturated. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 91.8 | #80 |
| 2 | Claude Opus 4.6 | 91.1 | #63 |
| 3 | Claude Opus 4.5 | 90.8 | #83 |
| 4 | Claude Opus 4.1 | 89.5 | #140 |
| 5 | Claude Sonnet 4.6 | 89.3 | #90 |
| 6 | Gemini 2.5 Pro | 89.2 | #152 |
| 7 | Claude Sonnet 4.5 | 89.1 | #142 |
| 8 | Claude Opus 4 | 88.8 | #163 |
| 9 | O1 | 87.7 | #183 |
| 10 | GPT-4.1 | 87.3 | #248 |
| 11 | Claude Sonnet 4 | 86.5 | #199 |
| 12 | Claude 3.7 Sonnet (Thinking) | 86.1 | #249 (Claude 3.7 Sonnet) |
| 13 | GPT-4.5 | 85.1 | #274 |
| 14 | Llama 4 Maverick | 84.6 | #464 |
| 15 | Claude 3.7 Sonnet | 83.2 | #249 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vellum-mmmlu · How It Works · Data refreshed daily, snapshot 2026-10-20.