ResearchBenchmarksAutonomous Agent Task Solving on AgentBoard AlfWorldFollow63.4Task Completion RateAdMem46.34450.77255.259.628Jun 5, 2026Evaluation ResultsMethodMethodLinksTask Completion RateAverage ProgressAdMemLLM backbone=Claude Ha...LLM backbone=Claude Haiku 4.52026.0663.477.55ReActLLM backbone=Claude Ha...LLM backbone=Claude Haiku 4.52026.0649.365.86AWMLLM backbone=Claude Ha...LLM backbone=Claude Haiku 4.52026.064754.67