1X Unveils World Model to Advance Robot Visual Learning

1X, the company behind the Neo humanoid robot, has introduced a new AI model designed to understand real-world dynamics and enable robots to learn independently.
This physics-based model, named the 1X World Model, utilizes video input and prompts to equip Neo robots with new capabilities. According to 1X, the system allows the robots to learn tasks beyond their original training.
The announcement coincides with 1X's preparations to launch its Neo humanoids for home use. The company began accepting pre-orders in October, with deliveries scheduled for this year. A company spokesperson declined to provide a specific shipping timeline or disclose order numbers, stating only that pre-orders had surpassed expectations.
"After years of developing our world model and refining Neo's human-like design, Neo can now learn from online video and directly apply that knowledge in the physical world," said Bernt Børnich, founder and CEO of 1X. "By translating prompts into new actions—even without prior examples—this represents the beginning of Neo's ability to self-teach and master a wide range of tasks."
The claim that the robot can transform any prompt into an action is ambitious and not entirely precise; for instance, you cannot simply instruct a Neo to drive a car and expect it to parallel park immediately. However, a learning process is indeed occurring.
A company spokesperson clarified that the current world model does not instantly enable Neo robots to perform a new task based solely on a video prompt. Instead, video data associated with specific prompts is processed by the model, and the insights are then distributed across the robot network. This enhances their collective understanding of the physical world and expands their practical knowledge.
The model also provides visibility into how Neo interprets and plans responses to given prompts. This behavioral data aids 1X in training the models, moving toward a future where robots can effectively respond to entirely novel instructions.
Join the Disrupt 2026 Waitlist
Secure your spot on the Disrupt 2026 waitlist for priority access to Early Bird tickets. Previous Disrupt events have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 industry experts leading more than 200 sessions designed to accelerate your growth and competitive advantage. Additionally, connect with hundreds of innovative startups across all sectors.
Join the Disrupt 2026 Waitlist
Secure your spot on the Disrupt 2026 waitlist for priority access to Early Bird tickets. Previous Disrupt events have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 industry experts leading more than 200 sessions designed to accelerate your growth and competitive advantage. Additionally, connect with hundreds of innovative startups across all sectors.
San Francisco|October 13-15, 2026WAITLIST NOW
Related article
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
California AV Compliance: A New Era of Tickets, Geofences, and 1M Miles
Guident operates an AuveTech shuttle in South Florida, managing a four-mile route in West Palm Beach and a one-mile route in Boca Raton using its remote monitoring technology. | Credit: GuidentCalifornia is redefining the regulatory landscape for dri
Related Special Topic Recommendations
Comments (2)
0/500
So they're basically giving robots a physics engine in their brains? That sounds cool but also kinda scary. How do they handle all the messy real-world noise that simulators can't capture? 🤔
Interesting approach, but I'm a bit skeptical about how well a physics-based world model can handle the messy unpredictability of real environments. The video input part sounds promising for learning, but I wonder about the computational cost and if it'll lead to robots that are good at specific tasks but lack common sense. The race for humanoid AI is heating up! 🤖

1X, the company behind the Neo humanoid robot, has introduced a new AI model designed to understand real-world dynamics and enable robots to learn independently.
This physics-based model, named the 1X World Model, utilizes video input and prompts to equip Neo robots with new capabilities. According to 1X, the system allows the robots to learn tasks beyond their original training.
The announcement coincides with 1X's preparations to launch its Neo humanoids for home use. The company began accepting pre-orders in October, with deliveries scheduled for this year. A company spokesperson declined to provide a specific shipping timeline or disclose order numbers, stating only that pre-orders had surpassed expectations.
"After years of developing our world model and refining Neo's human-like design, Neo can now learn from online video and directly apply that knowledge in the physical world," said Bernt Børnich, founder and CEO of 1X. "By translating prompts into new actions—even without prior examples—this represents the beginning of Neo's ability to self-teach and master a wide range of tasks."
The claim that the robot can transform any prompt into an action is ambitious and not entirely precise; for instance, you cannot simply instruct a Neo to drive a car and expect it to parallel park immediately. However, a learning process is indeed occurring.
A company spokesperson clarified that the current world model does not instantly enable Neo robots to perform a new task based solely on a video prompt. Instead, video data associated with specific prompts is processed by the model, and the insights are then distributed across the robot network. This enhances their collective understanding of the physical world and expands their practical knowledge.
The model also provides visibility into how Neo interprets and plans responses to given prompts. This behavioral data aids 1X in training the models, moving toward a future where robots can effectively respond to entirely novel instructions.
Join the Disrupt 2026 Waitlist
Secure your spot on the Disrupt 2026 waitlist for priority access to Early Bird tickets. Previous Disrupt events have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 industry experts leading more than 200 sessions designed to accelerate your growth and competitive advantage. Additionally, connect with hundreds of innovative startups across all sectors.
Join the Disrupt 2026 Waitlist
Secure your spot on the Disrupt 2026 waitlist for priority access to Early Bird tickets. Previous Disrupt events have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 industry experts leading more than 200 sessions designed to accelerate your growth and competitive advantage. Additionally, connect with hundreds of innovative startups across all sectors.
San Francisco|October 13-15, 2026WAITLIST NOW
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
So they're basically giving robots a physics engine in their brains? That sounds cool but also kinda scary. How do they handle all the messy real-world noise that simulators can't capture? 🤔
Interesting approach, but I'm a bit skeptical about how well a physics-based world model can handle the messy unpredictability of real environments. The video input part sounds promising for learning, but I wonder about the computational cost and if it'll lead to robots that are good at specific tasks but lack common sense. The race for humanoid AI is heating up! 🤖





Home






