Neo-humanoid maker 1X releases world model that allows bots to learn what they see

1X, the robotics company behind the Neo humanoid robot, has announced a new AI model that understands real-world dynamics and allows bots to independently learn new information.

This physics-based model, called the 1X World Model, uses a combination of videos and prompts to give your Neo robot new abilities. 1X says the video will allow the neobots to learn new tasks for which they have not been trained before.

This release comes as 1X prepares to release Neo Humanoid into homes. The company began taking pre-orders for the humanoid in October and plans to ship the bot this year. A 1X spokesperson declined to share a timeline for when these bots will ship or information on order numbers beyond saying pre-orders have exceeded expectations.

“After years of developing world models and making Neo’s design as human-like as possible, Neo can now learn from internet-scale video and apply that knowledge directly to the physical world,” 1X founder and CEO Bernt Børnich said in a statement. “With its ability to transform any prompt into a new action, even without precedent, this is the starting point for Neo’s ability to learn to master almost anything you can think of asking.”

It’s a lofty claim that a bot can turn any prompt into a new action, and it’s not entirely accurate. You can’t tell Neo to drive a car, and he’ll suddenly know how to parallel park, for example. But learning is happening.

1X is not saying its global model will allow today’s Neo bots to capture video and immediately perform new tasks upon prompting, a company spokesperson clarified. Instead, the bot retrieves video data linked to a specific prompt and sends it back to the world model. That model is fed back into a network of bots, providing a deeper understanding of the physical world and more know-how.

It also provides users with insight into how Neo intends to act or react to certain prompts. This kind of behavioral information could help train these models until the robot can respond to prompts it hasn’t done before.