BlueBench - Reasoning: leaderboard
BlueBench reasoning task combining commonsense and science QA challenges such as HellaSwag and OpenBookQA.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 | 79 | #183 |
| 2 | GPT-4o | 78 | #343 |
| 3 | GPT-4.1 | 77.5 | #248 |
| 4 | O4 Mini | 75.5 | #180 |
| 5 | Mistral Medium 3 | 75.5 | #500 |
| 6 | Mistral Large | 75 | #710 |
| 7 | Llama 3.3 70B Instruct | 73.5 | #534 |
| 8 | Pixtral-12B | 73.5 | #808 |
| 9 | GPT-4.1 Mini | 73 | #355 |
| 10 | O3 Mini | 73 | #272 |
| 11 | GPT-4.1 Nano | 69 | #731 |
| 12 | Granite 3.3 8B Instruct | 62 | #825 |
| 13 | Llama 3.2 3B Instruct | 56 | #1352 |
| 14 | Llama 3.2 1B Instruct | 41 | #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-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.