HELM MMLU - Professional Law: leaderboard

Metric: EM. Source: crfm.stanford.edu. 79 models tracked.

Top models

#ModelScore
1Claude 3.5 Sonnet (20240620)75.36
2Claude 3.5 Sonnet (20241022)74.71
3GPT-4o (2024-08-06)74.19
4GPT-4 (0613)72.69
5GPT-4o (2024-05-13)72.29
6Claude 3 Opus (20240229)71.51
7DeepSeek V371.32
8Gemini 1.5 Pro (002)70.8
9GPT-4 Turbo (Preview)70.66
10Llama 3.1 405B Instruct70.4
11Qwen 2.5 72B Instruct67.73
12Llama 3.2 90B Vision Instruct67.6
13Llama 3.1 70B Instruct67.41
14GPT-4 Turbo67.08
15Gemini 2.0 Flash (Preview)66.75

Interactive version: theaggregate.ai/benchmark?slug=helm-mmlu-professional-law · How It Works · Data refreshed daily, snapshot 2026-09-08.