Neo Humanoid Maker 1X Releases World Model to Help Robots Learn From What They See

The robotics company behind the Neo humanoid robot, 1X, has introduced a new artificial intelligence system designed to help robots better understand and learn from the real world.

The new physics-based system, called the 1X World Model, is built to interpret real-world dynamics and enable Neo robots to acquire new skills independently. By combining video input with written prompts, the model allows Neo to learn tasks it was not explicitly trained on, according to the company.

The announcement comes as 1X prepares to bring its Neo humanoid robots into homes. The company began accepting preorders for Neo in October and has said it plans to ship the robots later this year. A spokesperson for 1X declined to provide a specific shipping timeline or disclose the number of units ordered, noting only that demand has surpassed expectations.

“After years of developing our world model and making Neo’s design as close to human as possible, Neo can now learn from internet-scale video and apply that knowledge directly to the physical world,” said Bernt Børnich, founder and CEO of 1X, in a statement. “With the ability to transform any prompt into new actions — even without prior examples — this marks the beginning of Neo’s ability to teach itself to master nearly anything you could think to ask.”

While the company’s claims are ambitious, they are not without limits. For example, instructing a Neo robot to drive a car would not instantly enable it to parallel park. However, 1X states that the system supports meaningful learning.

According to a company spokesperson, the world model does not allow Neo robots to immediately perform a new task simply by watching a video and receiving a prompt. Instead, the robot collects video data for specific prompts and feeds it into the world model. The updated model is then distributed across the broader network of robots, improving their shared understanding of physical environments and task execution.

The system also provides insight into how Neo interprets and plans its behaviour in response to prompts. This visibility into the robot’s decision-making process could help 1X further train its models, with the long-term goal of enabling robots to respond effectively to tasks they have not previously performed.