NEO can now teach itself new skills using video-based AI models

1X has introduced a significant AI upgrade for its humanoid robot NEO, called the 1X World Model, which enables the robot to autonomously learn new physical skills from simple voice or text prompts—even for tasks and environments it has never encountered before. This breakthrough is powered by a video-based AI model grounded in real-world physics, allowing NEO to learn from vast internet-scale video data that captures human interactions with objects. Unlike traditional robots that rely heavily on pre-programmed behaviors or extensive human-operated data, NEO can generalize knowledge from videos and translate it into reliable physical actions, bridging a critical gap in humanoid robotics. The 1X World Model allows NEO to perceive its surroundings through cameras, generate visual predictions of future actions, and execute precise movements using an inverse dynamics model. Demonstrations show NEO performing everyday tasks such as packing a lunch box and handling novel actions like opening sliding doors, ironing clothes, and brushing hair without prior training examples. This capability reflects the transfer