← Benchmarks

Live leaderboard

MMLU

Massive Multitask Language Understanding, 5-shot. These rows are reported from a public leaderboard. Reported rows count at the 1.0× lane after human gate approval; Gaia has not independently reproduced them.

Benchmark ID mmlu@2024-03
Unit pct
Snapshot 2024-03-01
Methodology mmlu-v1
Last updated Loading…