List your company

Agent task performance on AgentBoard ScienceWorld Block Stream

64.4Success Rate (1-)ReAct
59.82461.01262.263.388Jun 1, 2026

Evaluation results

MethodLinks
ReAct2026.0664.487.164.487.1
ExpRAG2026.0664.487.865.686.9
ReMem2026.0664.485.762.285
MEMPROBE2026.066085.262.287.2