AILuminate Safety — leaderboard

Metric: Safety Grade (1-5). Source: ailuminate.mlcommons.org. 32 models tracked.

Top models

#ModelScore
1Claude 3.5 Sonnet (20241022)4
2Phi-44
3Claude 3.5 Haiku (20241022)4
4Gemma 2 9B4
5Phi-3.5-MoE-instruct4
6GPT-4o3
7GPT-4o Mini3
8Gemini 1.5 Pro3
9Gemini 2.0 Flash (001)3
10Llama 3.1 405B Instruct3
11Phi-3.5-mini-instruct3
12Gemini 2.0 Flash Lite3
13Mistral Large 2 (Nov) Instruct (2411)2

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