BlueBench - Knowledge: leaderboard
BlueBench knowledge task using MMLU-Pro subjects to test broad professional and academic knowledge with harder multiple-choice questions.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 | 71.43 | #183 |
| 2 | GPT-4.1 | 58.16 | #248 |
| 3 | Mistral Large | 55.1 | #710 |
| 4 | Llama 3.3 70B Instruct | 53.06 | #534 |
| 5 | O3 Mini | 50 | #272 |
| 6 | GPT-4o | 45.92 | #343 |
| 7 | GPT-4.1 Nano | 44.9 | #731 |
| 8 | Mistral Medium 3 | 44.9 | #500 |
| 9 | GPT-4.1 Mini | 42.86 | #355 |
| 10 | O4 Mini | 40.82 | #180 |
| 11 | Pixtral-12B | 38.78 | #808 |
| 12 | Llama 3.2 3B Instruct | 35.71 | #1352 |
| 13 | Granite 3.3 8B Instruct | 32.65 | #825 |
| 14 | Llama 3.2 1B Instruct | 19.39 | #1539 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=bluebench-knowledge · How It Works · Data refreshed daily, snapshot 2026-10-20.