Last updated: April 5, 2026 · Evaluation & Benchmarks · by Daniel Ashford
What is MMLU / MMLU-Pro?
A benchmark testing broad academic knowledge across 57 subjects.
Definition
MMLU (Massive Multitask Language Understanding) tests language models across 57 academic subjects including STEM, humanities, social sciences, and professional fields. MMLU-Pro is the harder updated version.
How It Works
The original MMLU contained 16,000+ multiple-choice questions. Frontier models have saturated it above 88%. MMLU-Pro addresses this with harder questions. Top MMLU-Pro scores in 2026 are around 79-81%.
Example
A sample MMLU question in professional law: "Under the Fourth Amendment, which of the following would most likely constitute an unreasonable search?" followed by four legal scenarios.
Related Terms
See How Models Compare
Understanding mmlu / mmlu-pro is important when choosing the right AI model. See how 12 models compare on our leaderboard.