An embodied AI robot powered by large models, flexibly designed for complex scenarios, created to better serve humanity. G1 is powered by a spatially intelligent ‘cerebrum’ large model. It understands 3D environments, interacts naturally with humans, and autonomously breaks down long-range tasks into actionable steps. G1 features a multi-tasking ‘cerebellum’ large model, trained on billions of simulated data. It delivers high-success-rate generalized embodied skills across diverse scenarios. Purpose-built for work, G1 combines wheeled mobility with a foldable lifting mechanism, ensuring smooth horizontal movement and a vertical reach of up to 2.4 meters, making it ideal for complex environments and diverse operational needs.
