Background
What is the MMLU?
MMLU stands for Massive Multitask Language Understanding. It is a broad multiple-choice benchmark used to compare how well people and language models handle general knowledge and reasoning questions across many domains.
What it covers
The benchmark spans 57 subjects, including mathematics, history, law, computer science, medicine, economics, and philosophy. This quiz samples from a local set of 14,042 questions derived from the MMLU test split.
Why the random baseline is 25%
MMLU questions are four-option multiple choice. If someone guessed every answer at random, they would average about 25% accuracy over a large number of questions.
Why compare with LLMs?
MMLU became a standard benchmark for tracking broad reasoning performance in large language models. This quiz shows your score alongside published benchmark numbers so you can place a short sample in that wider context.