ResearchBenchmarksLong-horizon tasks on Minecraft WoodFollow97.47Success Rate (SR)EvoAgent25.387644.101362.81581.5287Feb 9, 2025Evaluation ResultsMethodMethodLinksSuccess Rate (SR)Efficiency Error (EE)EvoAgent2025.0297.4798.43Optimus-12025.0296.3997.82LS-Imagine2025.0295.8797.41DreamerV32025.0291.0793.22Jarvis-12025.0289.7387.36GPT-4V2025.0235.2469.45PPO2025.0228.1653.82