#
# 1X Unveils World Model for Neo Bots to Learn from Visuals
In a groundbreaking advancement for humanoid robotics, 1X Technologies has launched the 1X World Model, empowering its NEO home robot to autonomously learn new skills from online videos and simple voice or text prompts, bridging the gap between digital AI and physical execution.[1][2]
Revolutionizing Robot Learning with Video-Based AI
The 1X World Model marks a paradigm shift by enabling NEO to analyze vast internet-scale videos of human-object interactions, grounded in real-world physics, to generalize behaviors in unfamiliar environments.[1][2] Unlike traditional robots reliant on human-operated data collection, NEO now teaches itself through a self-reinforcing cycle: it gathers data during interactions, predicts visual outcomes via its cameras, and executes actions using an inverse dynamics model.[1] 1X CEO Bernt Børnich emphasized, "NEO can now learn from vast online videos and apply that understanding directly to the physical realm," turning any prompt into new actions without prior examples.[1][2]
This video-centric approach allows NEO to handle dynamic, unpredictable home settings—like varying lighting or clutter—with human-like composure, a first in humanoid robotics.[2]
NEO's Vision-Only Design and Advanced Hardware
NEO employs a vision-only perception system with two 8-megapixel fisheye cameras offering 180° peripheral vision, inspired by Tesla's strategy to derive spatial understanding purely from visuals, ditching costly LiDAR sensors.[3] Powered by an Nvidia inference chip and 1X's Redwood AI—a generalist vision-language-action model—NEO processes natural language commands, navigates homes, and performs chores like opening doors or fetching items.[4][6][8] Early capabilities include companion mode for reminders and chore modes, with ongoing software updates enhancing autonomy as the robot learns from fleet data, accelerated by OpenAI's investment in compute infrastructure.[3][7]
Implications for Home Robotics and Scalability
The World Model scales NEO's intelligence alongside rapid video AI improvements, eliminating bottlenecks from slow human data gathering and positioning 1X for mass-market home deployment.[1][2] While initial tests showed some remote human oversight, the update pushes toward full autonomy, with pricing via early access purchase or subscription testing consumer demand.[5] As NEO ships in 2026, it promises to evolve from basic tasks to mastering "nearly anything you could think to request," potentially accelerating embodied AI toward general intelligence.[1][3]
Frequently Asked Questions
What is the 1X World Model? The 1X World Model is a video-based AI system that allows NEO robots to learn new tasks from online videos of human actions, predict outcomes, and execute them in real-world physics without prior training data.[1][2]
How does NEO learn from visuals? NEO uses its cameras to analyze videos, formulates visual forecasts for actions, and applies an inverse dynamics model to translate them into precise movements, generalizing across new environments.[1]
What hardware powers NEO's vision? NEO features two 8-megapixel fisheye cameras for 360° awareness, an Nvidia inference chip, microphones, speakers, and Redwood AI for vision-language-action processing—no LiDAR needed.[3][4][7]
Can NEO operate fully autonomously now? While the World Model enables self-learning, early demos included some remote assistance; it's designed for progressive autonomy through self-data collection and AI updates.[5][7]
When will NEO be available and what's the price? NEO is set to ship in 2026 with early access purchase or subscription options, priced around $20,000 in some reports, targeting home use.[5][7]
Who backs 1X and NEO's development? 1X has significant investment from OpenAI, providing compute power to speed up AI training from fleet data, betting on NEO as a platform for embodied AGI.[3]
🔄 Updated: 1/13/2026, 5:31:18 PM
**LIVE NEWS UPDATE: 1X World Model Launch Sparks Investor Buzz Amid Sparse Trading Data**
1X Technologies' unveiling of its **1X World Model** for NEO humanoid bots on January 12, 2026—enabling self-learning from internet-scale video data—drew positive analyst reactions, with experts hailing it as a "paradigm shift" that scales AI capabilities beyond slow robot data collection[1][3]. However, no immediate stock price movements were reported for the privately held firm, though pre-order details at **$20,000 early access** or **$499/month subscription** signal strong consumer market positioning for 2026 U.S. deliveries[2][5]. Rival humanoid developers like Tesla Optimu
🔄 Updated: 1/13/2026, 5:41:11 PM
**LIVE NEWS UPDATE: No Official Regulatory Response to 1X's World Model Launch for NEO Bots**
As of January 13, 2026, no government agencies or regulatory bodies have issued statements or actions regarding 1X Technologies' newly unveiled **1X World Model**, which enables NEO humanoid robots to learn skills from online videos and execute voice/text commands in novel environments[1]. Sources across tech outlets report zero mentions of oversight from bodies like the U.S. FTC, EU AI regulators, or Norway's DSA enforcers, despite NEO's $20,000 pre-order price and planned 2026 U.S. consumer rollout with privacy features like "no-go" zones[4]. Industry watchers note the absence of scrutin
🔄 Updated: 1/13/2026, 5:51:07 PM
**LIVE UPDATE: 1X's World Model Reshapes Humanoid Robotics Competition**
1X's new **1X World Model**—a 14-billion-parameter video system for NEO—breaks the data bottleneck plaguing rivals by leveraging internet-scale videos instead of slow human-operated collection, enabling self-improvement and execution in unseen environments like cluttered homes.[1][4] This "first of its kind" edge over traditional models, which struggle with lighting shifts or chaos, escalates pressure on competitors like Tesla's Optimus and Figure, as NEO readies for 2026 U.S. deliveries via purchase or subscription.[2][3][5] "This marks the starting point of Neo’s ability to teach itself t
🔄 Updated: 1/13/2026, 6:01:31 PM
**NEWS UPDATE: 1X World Model Reshapes Humanoid Race**
1X's new **World Model** for NEO humanoid robots—built on a **14B generative video model** trained on internet-scale data—breaks the data bottleneck plaguing rivals like Tesla's Optimus and Figure AI, which rely on slow human-operator collection, enabling NEO to self-learn tasks in unseen environments at "internet speed."[1][2][3] CEO Bernt Børnich stated, “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.”[1][2] Priced at **
🔄 Updated: 1/13/2026, 6:11:10 PM
**Breaking: 1X World Model Ushers in Self-Learning Era for NEO Humanoid Robots**
Industry experts hail 1X's new video-based world model as a "paradigm shift" in humanoid AI, enabling NEO to translate natural-language prompts into actions in unfamiliar environments using internet-scale video data fine-tuned on robot footage, bypassing slow human-operated data collection[1][2][4]. 1X CEO Bernt Børnich declared, “Neo can now learn from internet-scale video and apply that knowledge directly to the physical world... the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask,” while AI researcher Daniel Ho added, “With the 1X Worl
🔄 Updated: 1/13/2026, 6:21:10 PM
I cannot provide the market reactions and stock price movements you requested, as the search results do not contain any information about stock price data, investor reactions, or market impact following 1X's announcement of the world model for Neo robots.
The available sources focus exclusively on the technical capabilities of the 1X World Model—such as how it enables Neo to convert natural-language prompts into physical actions using video-based AI grounded in physics[1]—and mention that Neo will begin early-access delivery in 2026 through purchase and subscription options[1]. However, they do not include financial market data, trading activity, or analyst commentary that would be necessary for a complete breaking news update on market reactions.
To write
🔄 Updated: 1/13/2026, 6:31:13 PM
**BREAKING: 1X World Model Shakes Up Humanoid Robotics Race**
1X's new **1X World Model**, built on a **14B generative video model** trained with **900 hours of egocentric human video** and Neo-specific data, enables Neo bots to learn tasks from internet-scale videos rather than slow human-operator data collection—directly challenging rivals stuck on "vision language action" methods[1][4][6]. CEO Bernt Børnich declared, “this marks the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask,” positioning 1X ahead in dynamic home environments where traditional models falter amid lighting changes and clutter[1][2]. With Ne
🔄 Updated: 1/13/2026, 6:41:14 PM
**BREAKING: 1X World Model Marks Leap in Humanoid Learning.** Robotics experts hail 1X's new **1X World Model**—a **14B parameter generative video model** trained on **900 hours of egocentric human video** and fine-tuned on Neo sensorimotor data—as a paradigm shift from slow, operator-dependent training to internet-scale video learning, enabling Neo bots to predict and execute actions in dynamic home environments.[4][6] 1X CEO Bernt Børnich stated, “Neo can now learn from internet-scale video and apply that knowledge directly to the physical world... the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask,” while analyst
🔄 Updated: 1/13/2026, 6:51:12 PM
**LIVE NEWS UPDATE: Regulatory Response to 1X's World Model for Neo Bots**
No specific regulatory or government responses have emerged to 1X Technologies' unveiling of its new world model enabling Neo home robots to learn tasks from visuals alone, as announced on January 12, 2026.[3] In a Bloomberg Technology interview, 1X CEO and CTO Bernt Børnich stated, "We are doing... the proper things on the safety side with respect to following the standards, helping develop some of the standards, but also ensuring that we... have externally audited work of our safety work by an independent third party."[3] Officials have yet to comment publicly amid ongoing concerns over robot autonomy in homes, wit
🔄 Updated: 1/13/2026, 7:01:25 PM
**LIVE UPDATE: No Official Regulatory Response to 1X's World Model for Neo Bots**
As of January 12, 2026, no government agencies have issued statements or actions regarding 1X Technologies' new AI world model enabling Neo humanoid robots to learn tasks from video or voice prompts without prior experience[3]. In a Bloomberg Technology interview, 1X CEO and CTO Bernt Børnich addressed safety concerns, stating, "We are doing... the proper things on the safety side with respect to following the standards, helping develop some of the standards, but also ensuring that we... have externally audited work of our safety work by an independent third party"[3]. Industry observers note potential hurdles like privacy from huma
🔄 Updated: 1/13/2026, 7:11:07 PM
**BREAKING: 1X Unveils 1X World Model for Neo Humanoids, Enabling Video-Based Self-Learning**
Robotics firm 1X announced its **1X World Model** on January 13—a physics-grounded AI built on a **14B parameter generative video model**, trained on web-scale video, 900 hours of egocentric human footage, and Neo sensorimotor data—to let Neo bots predict and execute actions from text/voice prompts via imagined video rollouts[1][4][6]. CEO Bernt Børnich stated, “Neo can now learn from internet-scale video and apply that knowledge directly to the physical world... the starting point of Neo’s ability to teach itself to maste
🔄 Updated: 1/13/2026, 7:21:21 PM
**LIVE NEWS UPDATE: 1X World Model Sparks Global Robotics Race**
The unveiling of 1X's World Model for Neo humanoid robots, enabling self-learning from internet-scale video data, has ignited international excitement, with coverage from Norway-based 1X's U.S. headquarters rippling across outlets in Korea (Korea Bizwire), the UK (Robo Horizon), India (Bitcoin World), and the U.S. (TechCrunch, eWeek).[1][3][5][6] CEO Bernt Børnich declared it "the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask," positioning the $20,000 early-access bots for 2026 home deployment amid pre-order
🔄 Updated: 1/13/2026, 7:31:39 PM
**1X unveils physics-based world model enabling Neo robots to learn from video data**, positioning the company ahead of competitors relying on traditional demonstration-heavy training methods.[1][2] The breakthrough uses a **14-billion parameter generative video model** trained on internet-scale data and fine-tuned with egocentric human video, allowing Neo to perform untrained tasks by generating predicted future actions before executing them—a capability that CEO Bernt Børnich says marks "the starting point of Neo's ability to teach itself to master nearly anything."[4] This approach fundamentally shifts the competitive landscape by decoupling humanoid AI advancement from the bottleneck of operator-collected robot data, instead leveraging
🔄 Updated: 1/13/2026, 7:41:21 PM
**BREAKING: 1X Unveils 1X World Model (1XWM) for Neo Bots, Enabling Video-to-Action Learning**
The **1X World Model**, a physics-grounded video AI policy integrated into Neo humanoid robots, preprocesses 900 hours of egocentric human video for first-person manipulation alignment and an inverse dynamics model (IDM) trained on 400 hours of unfiltered robot data, allowing Neo to generate text-conditioned video forecasts from internet-scale footage and execute them via dual 8.85MP stereo cameras at 90Hz[1][4][5]. CEO Bernt Børnich stated, "Neo can now learn from internet-scale video and apply that knowledge directly to th
🔄 Updated: 1/13/2026, 7:51:25 PM
**BREAKING: 1X World Model Shakes Up Humanoid Robotics Race with Self-Learning Leap**
1X's new **1X World Model**, built on a **14B generative video model** trained on 900 hours of egocentric human video and Neo sensorimotor data, shifts the competitive landscape by enabling Neo bots to learn from internet-scale videos rather than slow human-operator data collection—surpassing traditional vision-language-action (VLA) models that struggle with physical dynamics.[1][4][6] This positions 1X ahead of rivals, as CEO Bernt Børnich stated: “Neo can now learn from internet-scale video and apply that knowledge directly to the physical world... the starting point of Ne