Kaggle Open Benchmarks - MMLU: leaderboard
Accuracy on MMLU, four-option multiple-choice questions across 57 subjects from elementary mathematics to law and medicine. Kaggle's standardized re-run on its own harness under the open-benchmarks organization, not the original authors' leaderboard; Kaggle's fraction is shown as a percentage (0-100, higher is better). Guessing scores about 25; frontier models are near the ceiling.
Metric: Accuracy (%). Source: www.kaggle.com. Saturation forecast: Estimated already saturated. 59 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon | 95.25 | #9 |
| 2 | Grok 4.5 | 94.54 | #54 |
| 3 | Gemini 3.7 Flash | 94.52 | #24 |
| 4 | Gemini 3.8 Flash | 94.42 | #30 |
| 5 | Gemini 3.6 Flash | 94.17 | #57 |
| 6 | Claude Opus 5 | 94.16 | #16 |
| 7 | GPT-6 Astra | 94.14 | #6 |
| 8 | Grok 4.6 | 93.97 | #29 |
| 9 | GPT-6 Sol | 93.94 | #19 |
| 10 | Gemini 3 Pro (Preview) | 93.86 | #71 |
| 11 | GPT-6.1 Sol | 93.83 | #7 |
| 12 | Gemini 3.5 Flash | 93.6 | #66 |
| 13 | GPT-5 | 93.48 | #105 |
| 14 | Claude Opus 4.1 | 93.39 | #146 |
| 15 | GPT-5.6 Terra | 93.3 | #40 |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-open-benchmarks-mmlu · How It Works · Data refreshed daily, snapshot 2026-11-05.