1X Unveils World Model to Teach Neo Robots New Tasks

1X, the company behind the Neo humanoid robot, released a new AI tool called the 1X World Model on Tuesday. This model helps Neo robots make sense of the real world and pick up new tasks on their own. The firm, based in Palo Alto, California, sees this as a key move toward getting its bots into homes, where they can handle chores like cleaning and fetching items. Pre-orders for Neo opened in October, with shipments set to start in the United States in 2026.

Background

1X started out focusing on robots that could work in offices and homes. Early on, the company built Eve, a wheeled robot that tested basic hardware and software ideas. That was back before 2023. From there, they moved to Neo, a full walking humanoid meant for everyday spaces.

Neo came in steps. First, there was Neo Beta in August 2024. Those units went to internal teams and a few early testers to check walking, grabbing objects, and staying safe around people. By September 2024, 1X shifted to Neo Gamma. They make these at a factory in Norway. Production picked up speed compared to Eve, letting them plan for thousands of units in 2025, then tens of thousands in 2026, and much more after that.

The company runs everything in-house, from parts to assembly. This keeps quality high and changes quick. They pick homes over factories for training because real living rooms and kitchens give messy, useful data. Neo weighs 66 pounds but can lift 150 pounds. Its body uses tendons for smooth, strong moves like a person. Soft covers hide sharp edges, so it is safe near kids or pets.

1X opened a waitlist for people wanting Neo. Early buyers will help test and teach the robots new tricks. The goal is a loop where users show tasks, bots learn, and everyone gets better over time.

Key Details

The new 1X World Model is a physics-based system. It takes videos and text prompts to teach Neo skills it did not have before. For example, show a clip of folding laundry, add a note, and the robot figures out how to do it in real life.

This builds on Redwood AI, 1X's main brain for Neo. Redwood handles seeing, talking, and moving all at once. It runs on the robot's own chip, with 160 million settings tuned from real runs with Eve and Neo. The model works at 5 frames a second, fast enough for chores like opening doors or climbing stairs.

Neo stands 5 feet 6 inches tall, about 66 pounds. Each hand has 22 ways to move for grabbing cups or clothes. The whole body has 75 move points. It walks at 1.4 meters per second normally, runs up to 6.2. It sits, kneels, or gets up using cameras for balance.

Safety comes first. A soft 3D lattice wraps the frame to cushion bumps. Joints block pinches. Noise stays under 22 decibels, quieter than a fridge. Hands handle water submersion, body takes splashes. The outer suit and shoes wash in a machine.

Production and Rollout

1X makes Neo Gamma in batches now. They aim for big numbers soon: thousands this year, tens of thousands next. Pre-orders beat what they expected, but no word on exact numbers or ship dates beyond early 2026 in the US.

A company spokesperson said units go to early access buyers first. More markets follow in 2027. Colors include tan, gray, and dark brown to fit homes.

“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,” Bernt Børnich, founder and CEO of 1X, said in a statement. “With the ability to transform any prompt into new actions — even without prior examples — this marks the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask.”

What This Means

This world model changes how robots learn. Before, they needed exact training for each job. Now, Neo watches a video or reads a prompt and tries the task. It predicts what happens next using physics rules, so moves stay real and safe.

Homes turn into training grounds. As more Neos ship, they gather data from daily life: spills to clean, toys to pick up, doors to open. That data feeds back to improve all bots. Users save time on chores, maybe 2 hours a day, the company says.

Other firms chase factory robots, but 1X bets on homes first. Messy spaces build smarter AI that works anywhere. Early users join a community to share tips and custom skills.

Challenges remain. Power lasts for a day of light work. Charging spots need planning. Teaching tricky tasks might call for a 1X expert via video link. Still, the model opens doors to broader use.

Competition heats up. Players like Boston Dynamics focus elsewhere, but 1X pushes consumer sales. Millions of units by decade's end could reshape homes. Robots handle basics, people focus on other things.

1X keeps the waitlist open. Buyers get updates on ships and features. The world model rollout fits their pace: hardware ready, AI growing, factories humming.