PassBench Leaderboard

last updated 2026-05-29
# Model AS Score ▼ G-Mean Speedup ▼ Correctness ▼ fast_1 ▼

AS Score is the headline metric — higher is better; Eager = 1.000 and TorchInductor = 0.706 are reference baselines. Hover a column header or see the paper for metric definitions.