The Maxwell embodied intelligence large model, developed by Wu Bin and Wang Shaojiang's team at the Institute of Industrial AI, Chinese Academy of Sciences, topped the MetaWorld benchmark with an overall success rate of 91.90% and a Hard-mode rate of 95.67%, surpassing models on the Evo-SOTA leaderboard. The 1B-parameter model supports single-GPU inference and edge deployment, and reached up to 99.1% success on some LIBERO tasks.