BlueBench - Safety — leaderboard

Metric: Score (%). Source: huggingface.co. 18 models tracked.

Top models

#ModelScore
1Mistral Large88.04
2GPT-4o87.84
3GPT-4.187.49
4GPT-4.1 Nano86.67
5GPT-4.1 Mini85.96
6Mistral Medium 384.68
7Llama 3.3 70B Instruct83.8
8O4 Mini83.55
9Llama 3.2 3B Instruct83.37
10Llama 3.2 1B Instruct83
11O182.96
12Pixtral-12B82.45
13O3 Mini80.61

Interactive version: theaggregate.ai/benchmark?slug=bluebench-safety · How the rankings work · Data refreshed daily, snapshot 2026-07-22.