AgiBot's general-purpose embodied base large model, GO-1, is a revolutionary AI model. Based on the innovative Vision-Language-Latent-Action (ViLLA) architecture, this model uses a multi-modal large model (VLM) and a Mixture-of-Experts (MoE) system to achieve efficient conversion from visual and language input to robot action execution. GO-1 can learn from human videos and real robot data, possesses strong generalization capabilities, and can quickly adapt to new tasks and environments with minimal or even zero data. Its main advantages include efficient learning ability, strong generalization performance, and adaptability to various robot bodies. The launch of this model marks a significant step towards the generalization, openness, and intelligence of embodied intelligence, and is expected to play an important role in commercial, industrial, and household applications.