#web agent evaluation

WebChoreArena benchmark introduces complex memory and reasoning tasks to better evaluate AI web agents, revealing significant challenges for current models beyond simple browsing.