What is true on World Models?

Who captures

Vangrid11AMD7World Labs6NVIDIA6Google5Generative Bionics3@chris_j_paxton3Runway3@smsehy3Anthropic2Ropedia2Skyfall AI2Apptronik2@michellelsun2Generalist2Odyssey2Tesla2@Scobleizer2Samsung1Schaeffler1Nebius1NXAI1Auki1Bonsai Robotics1Neura1DexForce1Fable1Princeton University1Veeda1@XRoboHub1GigaAI1AetherLab1Archon Robotics1Amazon1@leggedrobotics1Space Data1Physical Intelligence1OpenAI1Aether AI1@ricci_nov1Daimon Robotics1@oprydai1Axis Robotics1Caterpillar1@c_valenzuelab1Astribot1Figure1@PrismaXai1Dynamic Map Platform1Unitree1Momenta1Core Matter1@IlirAliu_1

Sourced numbers

Vangrid$9M“@vangrid_io raised a $9M seed round while the network was already being built in stealth and the product was already live.”@nguyenthambt on X

World Labs$100million“She raised roughly $100 million in 2024 at a valuation north of $1 billion”@stretchcloud on X

World Labs$1B“It just raised $1B earlier this year.”@TheHumanoidHub on X

← pulse .md twin

Nothing on this track

Published	Source	Entity	Captured intelligence	Topic
2:27 PM ET	@CyberRobooo	Samsung	@CyberRobooo2:27 PM ETSamsung @CyberRobooo on X Samsung is testing a humanoid designed for 24/7 operations. The bipedal robot is being trained for navigation and material handling at Samsung’s R&D center in Seoul, with staff and partners monitoring its operation in shifts. (see a bit in…	Humanoids
11:18 AM ET	@CyberRobooo	Anthropic	@CyberRobooo11:18 AM ETAnthropic @CyberRobooo on X Imagine AMRs moving parts through a factory while humanoid robots handle components or tend machines. In a shared workspace, a robot needs to make sense of more than what’s in front of it. It has to connect what it sees now with what it… @ropedia_ai· Oct 21, 2026 Ropedia Super Week, Day 3 \| S-Agent 2.0: Give a Coding Agent Spatial Perception, and It Beats GPT-6	Physical AI
11:00 AM ET	@SchaefflerGroup	Schaeffler	@SchaefflerGroup11:00 AM ETSchaeffler @SchaefflerGroup on X With the Lighthouse Factory in Taicang, Schaeffler has created something unique in its global network: a closed-loop manufacturing process in which humanoid robots assemble components for other humanoid robots. Find out more here.…	Humanoids
7:26 AM ET	@humanoidsdaily	Generative Bionics	@humanoidsdaily7:26 AM ETGenerative Bionics @humanoidsdaily on X First look, in renders, at Generative Bionics’ planned data factory: humanoids training in mock-ups of a workshop, a warehouse and a hospital room, each mirrored by a digital twin. Company tagline: “Starting in Europe. Built for the… @humanoidsdaily· Oct 21, 2026 Generative Bionics plans a 5,000 m² humanoid training hub in Italy, targeting Q4 2027. Powered by AMD Helios, it would combine real-world GENE.01 robot training with digital-twin simulations—the first of three planned facilities. https://t.co/EO812iStl5	Humanoids
7:01 AM ET	@AGkorthos	Generative Bionics	@AGkorthos7:01 AM ETGenerative Bionics @AGkorthos on X Generative Bionics plans a 5,000 m² Physical AI Data Factory in Italy, combining real-world humanoid training, digital twins and AMD Helios computing. The first of three planned facilities is targeted to open in Q4 2027.…	Humanoids
6:06 AM ET	@humanoidsdaily	Generative Bionics	@humanoidsdaily6:06 AM ETGenerative Bionics @humanoidsdaily on X Generative Bionics plans a 5,000 m² humanoid training hub in Italy, targeting Q4 2027. Powered by AMD Helios, it would combine real-world GENE.01 robot training with digital-twin simulations—the first of three planned facilities.…	Humanoids
10:57 AM ET	@TheHumanoidHub	Ropedia	@TheHumanoidHub10:57 AM ETRopedia @TheHumanoidHub on X Ropedia just released W0, an open world model trained on ~10,000 hours of egocentric human video. One set of weights predicts first-person views for humans, robot arms, and humanoids (Unitree G1, Fourier GR-1) using a shared camera +… @ropedia_ai· Oct 19, 2026 Ropedia Super Week, Day 1 \| Ropedia W0: Different Robots, One World Model	Humanoids
7:09 AM ET	@lukas_m_ziegler	Nebius	@lukas_m_ziegler7:09 AM ETNebius @lukas_m_ziegler on X Five days left on this one. 🏆 The @nebiusai Physical AI Awards, in partnership with @NVIDIARobotics, are open globally and close 25 October. I’m on the judging panel this year. Five categories, which double as a decent map of how the…	Physical AI
4:55 AM ET	@NEURARobotics	NXAI	@NEURARobotics4:55 AM ETNXAI @NEURARobotics on X A robot has to be able to remember, so we’re joining forces with NXAI to build out the memory of cognitive robots. Read the full story here: https://t.co/XgTz7X4P5J https://t.co/49OednrqUW	Physical AI
10:03 PM ET	@nguyenthambt	Vangrid	@nguyenthambt10:03 PM ETVangrid @nguyenthambt on X Building better AI models is only part of the challenge. Physical AI also needs accurate information about the environments where robots are expected to operate @vangrid_io . A dataset collected months ago may not reflect what exists at a…	Physical AI
2:22 PM ET	@spaceandtech_	Skyfall AI	@spaceandtech_2:22 PM ETSkyfall AI @spaceandtech_ on X What if AI could imagine and plan a complex design before building it? Skyfall AI, founded by deep learning pioneers behind Maluuba, is introducing SKY-001, the first engineering world model designed to turn text prompts or 2D sketches… @spisallyouneed· Oct 19, 2026 Introducing SKY-001, the first Engineering World Model. On CAD and BIM tasks, it beats GPT-6 Astra and every frontier model we tested: -> 51-125x faster - > 150-210x cheaper -> Higher IOU than all frontier models SKY-001 lets you dream a physical object before you build it. 18 months of work. Built by the team out of Maluuba's deep learning lab (acq. by @MSRResearch). The first ‘AI General Engineer’ will be the backbone for Autonomous Manufacturing.	Physical AI
2:07 PM ET	@oprydai	Skyfall AI	@oprydai2:07 PM ETSkyfall AI @oprydai on X engineering is one of the hardest problems for AI. generating a 3D object is one thing. understanding how thousands of components fit together, their mechanical constraints, and the sequence required to assemble them is an entirely… @spisallyouneed· Oct 19, 2026 Introducing SKY-001, the first Engineering World Model. On CAD and BIM tasks, it beats GPT-6 Astra and every frontier model we tested: -> 51-125x faster - > 150-210x cheaper -> Higher IOU than all frontier models SKY-001 lets you dream a physical object before you build it. 18 months of work. Built by the team out of Maluuba's deep learning lab (acq. by @MSRResearch). The first ‘AI General Engineer’ will be the backbone for Autonomous Manufacturing.	Physical AI
12:31 PM ET	@Scobleizer	Ropedia	@Scobleizer12:31 PM ETRopedia @Scobleizer on X Understanding the physical world is one thing. Predicting how it changes when you walk, reach, and grab is another. That’s where world models come in. For Day 1 of Ropedia Super Week, @ropedia_ai introduces Ropedia W0, a cross-embodiment… @ropedia_ai· Oct 19, 2026 Ropedia Super Week, Day 1 \| Ropedia W0: Different Robots, One World Model	Humanoids
4:00 AM ET	@Scobleizer	Auki	@Scobleizer4:00 AM ETAuki @Scobleizer on X Auki is worth parting attention to. When @TheHumanoidHub gets his video up I will share it. It is something I hope everyone in AI and robotics pays attention to. He explains why robots don’t need to take jobs better than anyone.… @sopranos_ak· Oct 17, 2026 This is what makes @Auki so interesting. Multiple robots and phones collaboratively build a shared digital twin of their environment. A robot detects a problem. A human investigates it through AR, leaves a spatial note, and assigns another robot to clean up. One shared spatial layer. Humans + robots working together. A spatial internet for robotics. 👌 https://t.co/LwW1bpOeBr	Physical AI
7:37 AM ET	@lukas_m_ziegler	AMD	@lukas_m_ziegler7:37 AM ETAMD @lukas_m_ziegler on X AMD just paid $8.2 billion to hire the godmother of AI 💸 @AMD is acquiring @theworldlabs, Fei-Fei Li’s spatial intelligence startup, in an all-stock deal. Li becomes AMD’s Chief Scientist, reporting directly to CEO Lisa Su. Two of the…	Physical AI
9:47 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:47 PM ETVangrid @nguyenthambt on X Recognizing an object is one thing. Understanding how it exists in physical space is another. @vangrid_io is working on a part of the Physical AI stack that deserves more attention: spatial data collected from real environments. A robot…	Physical AI
9:00 AM ET	@michellelsun	Bonsai Robotics	@michellelsun9:00 AM ETBonsai Robotics @michellelsun on X “The demo is just 1% of the business.” Tyler Niday, CEO of Bonsai Robotics, describes the work after a successful demo. Drawing on his work at Blue River and John Deere, Tyler explains Bonsai's approach to machine design and field…	Physical AI
8:51 AM ET	@WevolverApp	Apptronik	@WevolverApp8:51 AM ETApptronik @WevolverApp on X In an internal test, Google DeepMind robotics researchers instructed Gemini 4 Argon to watch a video of a human playing cornhole. Using its video understanding and advanced reasoning capabilities, Gemini 4 built a virtual simulation in…	Humanoids
4:13 AM ET	@NEURARobotics	Neura	@NEURARobotics4:13 AM ETNeura @NEURARobotics on X When a robot opens a lock, only sound and touch can tell it whether the job is done. That’s why Neuron AI, our Perception-Language-Action (PLA) model, combines several senses at once. 3D images, tactile and force signals, audio, the…	Physical AI
9:27 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:27 PM ETVangrid @nguyenthambt on X Some of the most valuable data for Physical AI may come from places people usually overlook. @vangrid_io makes this problem tangible through its live capture bounty board. Requests include damaged locker doors, CCTV columns, ATM lobbies,…	Physical AI
1:18 PM ET	@theworldlabs	World Labs	@theworldlabs1:18 PM ETWorld Labs @theworldlabs on X Atlas beta access has started rolling out! Here’s a look at how you can control camera paths with pixel-perfect precision. https://t.co/lvmSp8aU4R	Physical AI
11:36 AM ET	@pstAsiatech	DexForce	@pstAsiatech11:36 AM ETDexForce @pstAsiatech on X Chinese startup DexForce trains humanoid robots using virtual data to slash costs https://t.co/Es0xaApu4r	Humanoids
9:01 AM ET	@smsehy	Google	@smsehy9:01 AM ETGoogle @smsehy on X Model progress in high-level task planning is moving at breakneck speed. The real moat shifts toward the engineers who understand physical packaging, deterministic RTOS boundaries, and plant floor integration. https://t.co/WcX34lFz9I @Zergylord· Oct 13, 2026 Argon is so good at robotics that our whole team is having an existential crisis right now, enjoy! https://t.co/jYc4Q4bhAI	Physical AI
5:56 AM ET	@smsehy	NVIDIA	@smsehy5:56 AM ETNVIDIA @smsehy on X Field agriculture is the purest sim-to-real stress test. Synthetic data creates the baseline envelope, but physical dust, changing soil moisture, and direct solar glare force edge compute to adapt in real time. https://t.co/NtTRjpXKQ8 @NVIDIARobotics· Oct 9, 2026 Where physical AI meets the farm 🌱 @Aigeninc is using NVIDIA Cosmos and NVIDIA Jetson to build solar-powered robots for autonomous, chemical-free weed control, reducing reliance on herbicides and diesel-powered field passes. With NVIDIA technology, Aigen can: ✅ Train perception models with 99% synthetic data ✅ Reduce data collection and labelling to turn out effective models in hours instead of seasons Read the story: https://t.co/bJrwaRS3mX	Physical AI
5:06 AM ET	@smsehy	Google	@smsehy5:06 AM ETGoogle @smsehy on X Software models adapt in minutes inside simulation. On the factory floor, the hurdle is translating zero-shot policy discovery into deterministic cycle times that survive 80,000 hours of continuous vibration. https://t.co/pn3sobvFfu @keerthanpg· Oct 13, 2026 Gemini 4 Argon is a step change and will have massive impact on robotics. We have published a developer blog to show various ways in which Argon accelerates robotics. https://t.co/8uHCuEizsX https://t.co/sO1Tpe1LIQ	Physical AI
10:57 PM ET	@kyleichan	Fable	@kyleichan10:57 PM ETFable @kyleichan on X The big question in China’s robotics industry: “Will general AI models eat up robotics? Many in China’s robotics industry are watching closely as US general-purpose models like Astra, Fable, and Gemini 4 Argon show impressive robotics… @latepostnews· Oct 15, 2026 2026 年 Q3，具身智能领域最有冲击力的变化，来自一个通用模型：9 月初 OpenAI 发布 GPT-6 Astra 后，具身从业者们发现，Astra 不只理解任务、拆解任务，还能输出机器人末端执行器的位姿，完成了本该属于 VLA 等物理 AI 模型的部分工作，而且还做得很不错。在部分测试中，直接把 Astra 当做策略模型（VLA、WAM 等都是机器人上的策略模型类型）的成功率已超过了一些专门为机器人任务和物理世界操作训练的具身模型。这动摇了数字世界 AI 和物理 AI 的边界，也引发了一个灵魂拷问： GPT、Fable 之类的通用模型是否会吃掉 Robotics？同时已经发生的事实是：OpenAI、Anthropic、DeepSeek、Kimi 等多数通用前沿模型公司都正酝酿进入 robotics。除了 Astra 冲击外，本期具身季报还关注以下 4 个行业进展和事件： - AMD 拟以 82 亿美元收购 World Labs - 宇树 IPO 及连锁反应 - 有可能大幅压缩部署成本的 “上下文学习” - 美国 FCC 实际禁止了中国机器人进口对一些人来说，Astra 的变化已经撼动了独立的具身智能公司的发展叙事和估值逻辑。至少，现在到了重新思考物理 AI 与通用 AI 边界的时刻。 https://t.co/ZRNrzkIDMe	Physical AI
9:23 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:23 PM ETVangrid @nguyenthambt on X Building infrastructure for Physical AI requires more than understanding the technology. It requires knowing how to turn a promising idea into a system that can operate at scale. @vangrid_io brings together two complementary backgrounds.…	Physical AI
3:34 PM ET	@johnkoetsier	Apptronik	@johnkoetsier3:34 PM ETApptronik @johnkoetsier on X Insane! Frontier models are revolutionizing humanoid robot training. Gemini 4 Argon ... - watched 1 video of a human playing cornhole - built its own sim - practiced - got Apptronik’s Apollo to sink 7 straight And it's just the beginning…	Humanoids
3:06 PM ET	@lukas_m_ziegler	Princeton University	@lukas_m_ziegler3:06 PM ETPrinceton University @lukas_m_ziegler on X Princeton's Introduction to Robotics! 🎓 Princeton University released their full Introduction to Robotics course publicly with lecture videos, notes, slides, and assignments. This course provides fundamental theoretical and algorithmic…	Physical AI
12:52 PM ET	@Ken_Goldberg	Anthropic	@Ken_Goldberg12:52 PM ETAnthropic @Ken_Goldberg on X Excited to release the RoboSimStudio toolkit today to help agents create detailed dynamics Real2Sim for Agentic Robotics. https://t.co/C43Mpx1v6y @keplerccccc· Oct 14, 2026 We gave Claude and GPT a phone video of a robot wiping. In an hour, each built a full-dynamics digital twin. Introducing RoboSimStudio 🤖: an open and interactive toolkit for Agentic Robotics, enabling Fable, Astra and Argon AI agents to model sponges 🧽, cables 🧵 and cloth 👔. https://t.co/BXf5sLXq51	Physical AI
10:24 AM ET	@michellelsun	Google	@michellelsun10:24 AM ETGoogle @michellelsun on X Google released Gemini 4 Argon yesterday. It scores 87% on ERQA, the embodied reasoning benchmark, first among frontier models. What stood out to me is Real2Sim. How it works: Argon takes one photo of a workspace. It builds a simulation…	Physical AI
10:06 AM ET	@chris_j_paxton		@chris_j_paxton10:06 AM ET @chris_j_paxton on X But will a 100x larger model do it https://t.co/nhDLTyglvc @JiahuaDong· Oct 12, 2026 [1/7] We trained a world model on 30,000 hours of first-person video. It learned exactly where the hands go. What happens to the things they touch, much less so. Going from 300 to 30,000 hours did not narrow the gap, and a 4× larger model barely did either. https://t.co/2EBqLD1L0c	Physical AI
8:03 AM ET	@michellelsun		@michellelsun8:03 AM ET @michellelsun on X If you’re building in robotics/physical AI, share one line about what you're working on. What's one specific problem your team is trying to solve? or wish someone will solve? https://t.co/mcNCMgfcfp @michellelsun· Sep 29, 2026 if you are building: - actuators - dexterous hands - training data for physical ai - self-driving labs - magnets - the deployment layer of robotics - reindustrialization - space manufacturing - energy and battery innovation - high-fidelity, physics-based world models - BCI for robot control - open-source frontier robot foundation models - anything frontier and sci-fi in physical AI welcome. this account is for you. tell me what you're working on 👇	Physical AI
6:49 AM ET	@lukas_m_ziegler	Veeda	@lukas_m_ziegler6:49 AM ETVeeda @lukas_m_ziegler on X one of the hottest places to work on solving physical AI hardest problems! also a rockstar team @FidlerSanja, @ZGojcic, @HuanLing6 🔥🦾 https://t.co/7wfp3s6z3c @FidlerSanja· Oct 13, 2026 At @VeedaAI , we’re building for a future where robots learn by interacting with the world, at scale -- in simulated reality. Getting there means bringing together the brightest minds to tackle some of the hardest problems in Physical AI. Come build what’s next with us: https://t.co/XFDUqnmbDm	Physical AI
2:31 AM ET	@XRoboHub		@XRoboHub2:31 AM ET @XRoboHub on X This robot is making wontons. One hand cups the wrapper while the other adds the filling, then both hands fold it closed. It’s running Natura-1, @naturewill’s 50B-parameter mixture-of-experts model. The team says it was pretrained on… @naturewill· Oct 13, 2026 Introducing Natura-1, our first general foundation model for the physical world. It follows natural-language instructions zero-shot and turns them into action with dexterous hands. Natura-1 is a 50-billion-parameter mixture-of-experts (MoE) model pretrained on one million hours of egocentric human experience. On top of this foundation, we conduct efficient post-training on a limited amount of robot data to adapt pretrained world knowledge to dexterous manipulation, yielding a model that generalizes to a wide range of unseen scenarios. Natura-1 follows complex natural-language instructions zero-shot, reasoning through spatial constraints, multi-object relationships, and negative conditions. It can also autonomously carry out long and continuous workflows to perform long-horizon executions. Natura-1 provides early evidence that generalization—and even emergent behavior—can coexist with the complexity of dexterous hands. Further progress depends on the full stack: infrastructure for faster training and real-time inference, pretraining data at greater scale and coverage, and models that transfer to robots with less adaptation. We expect it to be a long-term research and engineering effort.	Physical AI
6:44 PM ET	@ZappyZappy7	Google DeepMind	@ZappyZappy76:44 PM ETGoogle DeepMind @ZappyZappy7 on X 『DeepMindのGemini 4 Argon、人の動画をロボットの練習環境へ』 https://t.co/qz6Gn0aJzR 袋を板の穴へ投げる映像から空間と全身動作を推定し、物理シミュレーションを自動生成。人型ロボットの実験で7投連続成功。動画⇒仮想環境⇒実機が新技能の開発負担を減らす可能性。 #PhysicalAI #Sim2Real https://t.co/I0fqAYFBut	Humanoids
4:12 PM ET	@wolfejosh	Runway	@wolfejosh4:12 PM ETRunway @wolfejosh on X EXCELLENT from @packyM on Lux family co @runwayml https://t.co/7xbU9D90Cz @packyM· Oct 13, 2026 Runway has been on a tear over the past couple months. Viral HORSE Video contest Solaris: the world's first Interface World Model Praxis-1: open-weight World Action Model for robotics Project Continuum: operating system research application built around real-time interfaces A video model. An interface model. A robotics model. An operating system. Are you getting it? These are not separate pretrains. They’re different applications, built around the same General World Model backbone, with different post-training on top. All of them are examples of what you can do when you can simulate any pixel. So I asked co-CEO Anastasis Germanidis to co-write an essay with me explaining how video, interfaces, operating systems, and robotics all fit together, why video was the right starting point for World Models, where the world is heading if Runway’s bets are right, and whether or not, as someone who simulates worlds, he believes we’re living in a simulation. Welcome to the future. Enjoy! (Video, of course, made with Runway)	Physical AI
3:42 PM ET	@pstAsiatech	GigaAI	@pstAsiatech3:42 PM ETGigaAI @pstAsiatech on X China’s GigaAI Opens General World Model Lab in Beijing L5 aims is to understand the physical world in an open environment and achieve continuous interaction and dynamic evolution of multi-agent systems. https://t.co/4n8xooXbdy	Physical AI
2:07 PM ET	@IlirAliu_	AetherLab	@IlirAliu_2:07 PM ETAetherLab @IlirAliu_ on X Causality makes robots smarter! A robot that can handle everyday household tasks more safely and reliably by thinking through what will actually happen before it acts. Instead of just copying patterns, the system predicts the consequences…	Physical AI
9:14 AM ET	@chris_j_paxton		@chris_j_paxton9:14 AM ET @chris_j_paxton on X This has been exactly my feeling about what video models will do. Fundamentally there is a lot of missing information still. Even humans -- our brains and bodies honed by millions of years of evolutionary data for manipulation -- cant like… @JiahuaDong· Oct 12, 2026 [1/7] We trained a world model on 30,000 hours of first-person video. It learned exactly where the hands go. What happens to the things they touch, much less so. Going from 300 to 30,000 hours did not narrow the gap, and a 4× larger model barely did either. https://t.co/2EBqLD1L0c	Physical AI
9:38 AM ET	@chris_j_paxton	Archon Robotics	@chris_j_paxton9:38 AM ETArchon Robotics @chris_j_paxton on X Standing the bike back up is a cool example. Very optimistic about humanoid robots. https://t.co/rDmbqMwikb @Archon_Robotics· Oct 10, 2026 🤖 Building Whole-Body Intelligence for the Real World. Were Just Getting Started. From bending down to pick up objects and navigating steps to autonomously clearing tables, operating dishwashers, and organizing refrigerators, our robots are learning to perceive, move, manipulate, and coordinate their entire bodies in real-world environments. Powered by our Whole-Body Intelligence (WBI) framework, were building generalist humanoid foundation models by combining large-scale human whole-body data pretraining, context-aware behavioral foundation models, and efficient post-training with real-robot data - turning human experience into real-world robotic skills. 👉 Watch Whole-Body Intelligence in action in our three-minute demo. More to come. Stay tuned at https://t.co/XSoqTfuZaD #ArchonRobotics #WholeBodyIntelligence #WBI #Humanoid #PhysicalAGI	Humanoids
9:57 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:57 PM ETVangrid @nguyenthambt on X Collecting more spatial data is only part of the challenge. Making sure every capture meets the required standard is what makes the data useful. @vangrid_io approaches this problem with a workflow that connects capture requests, quality…	Physical AI
7:00 PM ET	@stretchcloud	Amazon	@stretchcloud7:00 PM ETAmazon @stretchcloud on X Decart is reportedly closing in on a $7 billion sale to Amazon, after Anthropic offered $6 billion in August and SpaceX reportedly looked too. I keep coming back to what's actually being bought here. Decart doesn't make a chatbot. It makes… @Polymarket· Oct 11, 2026 JUST IN: Amazon reportedly in talks to acquire AI startup Decart for approximately $7,000,000,000.00.	Physical AI
12:24 AM ET	@leggedrobotics		@leggedrobotics12:24 AM ET @leggedrobotics on X In our latest work at @corl_conf, we introduce Quadrupedal World Model: a single morphology-conditioned dynamics model, where policies are trained entirely in imagination. We found generalization emerges by conditioning the dynamics, not… @mo_danesh· Oct 7, 2026 What if one world model could control many different robots? 🤖🌍 That's QWM. In this 2 minutes video we show keypoints about QWM: learning across heterogeneous quadrupeds, and transferring zero-shot to unseen morphologies. No fine-tuning. No warm-up. No retraining. Let's talk it further at CoRL 🥳	Humanoids
9:34 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:34 PM ETVangrid @nguyenthambt on X The next challenge for Physical AI is building a trustworthy data pipeline, not just collecting more footage. @vangrid_io is developing spatial infrastructure that brings edge computed privacy, cryptographic provenance, multi view…	Physical AI
5:03 PM ET	@NVIDIARobotics	NVIDIA	@NVIDIARobotics5:03 PM ETNVIDIA @NVIDIARobotics on X Where physical AI meets the farm 🌱 @Aigeninc is using NVIDIA Cosmos and NVIDIA Jetson to build solar-powered robots for autonomous, chemical-free weed control, reducing reliance on herbicides and diesel-powered field passes. With NVIDIA…	Physical AI
3:12 PM ET	@TheHumanoidHub	Google DeepMind	@TheHumanoidHub3:12 PM ETGoogle DeepMind @TheHumanoidHub on X "What does a Post-AGI physical world look like?" That's @xiao_ted's keynote at the Humanoid Hub Conference. He co-authored major milestone projects such as SayCan, RT-1, RT-2 and Open X-Embodiment, and later Gemini Robotics. Now building…	Humanoids
9:37 AM ET	@ZappyZappy7	Generalist	@ZappyZappy79:37 AM ETGeneralist @ZappyZappy7 on X GeneralistのGEN-1.5、数秒の実演(お手本)から追加学習なしで新しい仕事に適応。現場ごとの作業立ち上げ負担を減らす可能性。 https://t.co/jmKZHtoWEX #PhysicalAI #RobotLearning #foundation #model #ZeroShot #Generalist https://t.co/diw0EvQQU3	Physical AI
4:56 AM ET	@logi_today	Space Data	@logi_today4:56 AM ETSpace Data @logi_today on X 倉庫を最短即日で3D化、AMR走行を事前検証-https://t.co/GHGGMLYZFv スペースデータは、実在する物流倉庫を最短即日で3D仮想空間化する手法を実証。AMRやフォークリフトの走行、荷物との干渉、複数台稼働時の動線を、ロボット導入前に検証できる。 https://t.co/7O1v9VQCNc	Physical AI
4:49 AM ET	@CyberRobooo	Physical Intelligence	@CyberRobooo4:49 AM ETPhysical Intelligence @CyberRobooo on X A humanoid robot that can catch a moving cup Most robots freeze the second something starts moving. Cup on a belt, shifting grasp, anything not stuck in place --and they miss. Long-WAM from NVIDIA (with MIT, HKU, UCSD) went after that. On…	Humanoids
6:52 PM ET	@johnkoetsier	Generalist	@johnkoetsier6:52 PM ETGeneralist @johnkoetsier on X Generalist physical AI? 1 AI brain, 6 domains ... Odyssey-3 just demonstrated a single world model controlling: 🤖 Humanoids 🦾 Robot arms 🚗 Self-driving cars 🚁 Drones 🎮 Video games 🌎 AI training environments This is pretty huge if it…	Physical AI
6:52 PM ET	@NVIDIARobotics	OpenAI	@NVIDIARobotics6:52 PM ETOpenAI @NVIDIARobotics on X Turning a simulation idea into a working application takes a lot of work. Frontier AI agents are making it easier. 👏 Developers are combining models like @OpenAI GPT-6 Astra with NVIDIA Omniverse libraries to prepare scenes, connect…	Physical AI
5:55 PM ET	@stretchcloud	Odyssey	@stretchcloud5:55 PM ETOdyssey @stretchcloud on X Odyssey trained one frozen model to drive cars, walk humanoids, fly drones, and run game worlds. Then it built a separate lightweight decoder for each job on top of that single backbone. The model shipped today as Odyssey-3, a free public… @AGTPinsights· Oct 8, 2026 Odyssey just launched Odyssey-3, a foundation world model, today. Here's what you need to know. Odyssey calls Odyssey-3 its most powerful foundation world model yet. It's free to try today as a research preview open to the public. The company says the same model powers robots, trains AIs, and generates interactive experiences. Odyssey-3 Pro set a new state of the art on Physics-IQ Verified's video-to-video benchmark. It scored 66.1, the highest reported score. Physics-IQ is a DeepMind benchmark. It tests physical behavior across fluid dynamics, optics, solid mechanics, magnetism, and thermodynamics. One post says the 66.1 is a best-of-8 result. It says a single run with enhanced prompts gets 63.4, and image-to-video tops out at 54.7. The model builds an interactive environment from a prompt, all in real time. It learns physics, dynamics, and cause-and-effect from a broad dataset of visual observations. Users can move in first or third person, steer the camera on its own, and trigger events. The scene then responds live. In one demo, a seal is controlled with WASD keys. It also works beyond games. Odyssey says the model has shown it can control robots and humanoids, drive vehicles, and pilot drones. It can also generate environments where AI agents train. In total, that covers six domains: robotics, humanoids, autonomous vehicles, AI training, drones, and gaming. One post says it learned to drive on real roads from just 20 hours of driving data. Odyssey's pitch is that physics is to world models what coding is to language models. If a model can predict how things move and react, robots and agents can work out what happens next. The team is especially excited about agents learning from experience inside the model, working toward objectives. One example is an agent told to "pick up the green cube" that learns by trying, instead of making costly mistakes in the real world. Key numbers: - 66.1: Odyssey-3 Pro on Physics-IQ Verified video-to-video, the highest reported - 63.4: single run with enhanced prompts - 54.7: image-to-video top score - 20 hours: driving data reportedly used to learn road driving - 6 domains: robots, humanoids, vehicles, AI training, drones, gaming Odyssey-3 is live today, free to try, as a single world model for physical AI, AI training, and interactive worlds.	Physical AI
3:16 PM ET	@humanoidsdaily	Aether AI	@humanoidsdaily3:16 PM ETAether AI @humanoidsdaily on X “What surprised us most was how different ‘the action completed’ is from ‘the intended outcome actually happened.’” Aether AI told Humanoids Daily what it learned putting its causal world model on a real robot for CRIS-0, and where the…	Humanoids
1:22 PM ET	@lukas_m_ziegler	Odyssey	@lukas_m_ziegler1:22 PM ETOdyssey @lukas_m_ziegler on X a toddler simulator that helps robots learn about our world wasn’t on my 2026 physical AI bingo…. https://t.co/hEQzp1FWf0 @odysseyml· Oct 8, 2026 Odyssey-3 can generate interactive environments from a prompt, all in real time. It's a new kind of world simulator, and learns representations of physics, dynamics, and cause-and-effect from a broad dataset of visual observations. https://t.co/FhDVYZVKuE	Physical AI
9:29 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:29 PM ETVangrid @nguyenthambt on X Physical AI will not scale on models alone @vangrid_io is interesting because the missing piece may be coordination between the physical world and the systems learning from it. Real places constantly change, but useful spatial data needs…	Physical AI
7:11 AM ET	@smsehy		@smsehy7:11 AM ET @smsehy on X LeCun points directly at the industrial reality. Language models predict tokens in a latent space, but machines on a plant floor or vehicles on tarmac must handle contact dynamics and unmodeled friction. Physical AI requires world models… @rohanpaul_ai· Oct 5, 2026 Yann LeCun's latest talk at ETH Zürich "You should not work on LLM. At least if you're in academia, you should absolutely not work on LLMs. There is nothing you can bring to the table. If you're interested in making real progress in AI, in sort of grounded AI for the real world, if you want physical AI, don't work on LLMs, and don't work on generative models either. So, as you can probably guess, this does not make me very popular in Silicon Valley." ---- From "Perfology Clips" YouTube channel, (link in comment)	Physical AI
9:18 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:18 PM ETVangrid @nguyenthambt on X The missing layer in Physical AI may be coordination @vangrid_io is interesting because collecting spatial data is only one part of the problem. You also need a network that can coordinate people, locations, data requests and verification…	Physical AI
5:47 PM ET	@ricci_nov		@ricci_nov5:47 PM ET @ricci_nov on X RealtimeWAM cuts world-action-model inference to one step and overlaps the video and action experts through a shared KV cache: ~25x end-to-end speedup on an H100 with under 1% performance drop. Robot inference turns into a scheduling…	Physical AI
2:00 PM ET	@NVIDIARobotics	NVIDIA	@NVIDIARobotics2:00 PM ETNVIDIA @NVIDIARobotics on X Closing the Real-to-Sim-to-Real Loop for Physical AI https://t.co/lVZ65Exh5I	Physical AI
9:54 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:54 PM ETVangrid @nguyenthambt on X Physical AI has a data problem that bigger models cannot solve alone @vangrid_io is interesting because the answer is not simply collecting more of the same data A robot or world model may need information from a specific street, building,…	Physical AI
12:13 PM ET	@techniahqrobot	AMD	@techniahqrobot12:13 PM ETAMD @techniahqrobot on X NEWS AMD is paying $8.2 billion for World Labs and the hardware angle is hard to ignore. World Labs builds spatial intelligence models that generate reconstruct and simulate interactive 3D environments from text images and video. Its…	Physical AI
9:38 PM ET	@nguyenthambt	Vangrid	@nguyenthambt9:38 PM ETVangrid @nguyenthambt on X The interesting part about Vangrid is what happened before the funding announcement @vangrid_io raised a $9M seed round while the network was already being built in stealth and the product was already live. The round brought together…	Physical AI
8:05 PM ET	@smsehy		@smsehy8:05 PM ET @smsehy on X Unseen environments break discrete task execution. Embodied agents require continuous spatial intelligence to track shifting geometry and physical affordances in real time. Solving long-horizon tasks without environment-specific retraining…	Physical AI
9:59 AM ET	@CyberRobooo	Daimon Robotics	@CyberRobooo9:59 AM ETDaimon Robotics @CyberRobooo on X Robots can now work by feel. One interesting demo at IROS 2026 comes from Daimon Robotics: a robot picks up tiny beads, aligns their holes, and threads them onto a flexible string. It looks simple, but the tricky part starts when the robot…	Physical AI
8:21 AM ET	@CyberRobooo	Runway	@CyberRobooo8:21 AM ETRunway @CyberRobooo on X Runway just released Praxis-1, a World Action Model for robots. It comes after GWM-1, the world model Runway introduced in December 2025. GWM-1 focused on predicting how the world changes in response to robot actions, opening up…	Physical AI
4:24 PM ET	@oprydai		@oprydai4:24 PM ET @oprydai on X it’s honestly insane how many pieces of the robotics stack are finally becoming good enough at the same time we now have - cheap high torque actuators - compact motors and gearboxes - better batteries - force and tactile sensing - depth…	Humanoids
1:51 PM ET	@NVIDIARobotics	NVIDIA	@NVIDIARobotics1:51 PM ETNVIDIA @NVIDIARobotics on X What will it take for AI to move beyond generating pixels and start understanding worlds? At #NVIDIAGTC Berlin, hear from @bfl_ai, @ltx_io, and @vesoma_ai for a conversation about the future of world models. Explore whether traditional…	Physical AI
2:18 AM ET	@smsehy		@smsehy2:18 AM ET @smsehy on X A true 3D world model needs more than pixels. It needs interactive environments grounded in physical reality and physics-ready simulation assets to build spatial intelligence. https://t.co/8veQoFHhua	Physical AI
11:00 PM ET	@axisrobotics	Axis Robotics	@axisrobotics11:00 PM ETAxis Robotics @axisrobotics on X Announcing our partnership with @ManycoreTech Manycore Tech is a leading global provider of spatial intelligence, focused on accelerating the integration of AI into the physical world. Axis Suite delivered a set of complex, physics-ready…	Physical AI
7:41 PM ET	@IlirAliu_	Caterpillar	@IlirAliu_7:41 PM ETCaterpillar @IlirAliu_ on X A robot arm can now be trained over a weekend by three students in San Francisco. A bulldozer cannot. Spent part of the day on the floor at Fully Connected, @CoreWeave's conference in San Francisco. Good energy, plenty of cool demos. My…	Physical AI
2:54 PM ET	@c_valenzuelab		@c_valenzuelab2:54 PM ET @c_valenzuelab on X Today we're announcing Praxis-1, an open-weight world action model that adapts to new robot embodiments with light finetuning. Video is eating the world https://t.co/yDhA9nxcOG	Physical AI
7:36 AM ET	@CyberRobooo	Astribot	@CyberRobooo7:36 AM ETAstribot @CyberRobooo on X Humanoid robots are starting to work together. Astribot is bringing T1(>$18K) to the US for the first time at IROS 2026, showcasing autonomous tasks like backpack packing,lab teleoperation...alongside Astribot’s in-house AI model,--Lumo-2.…	Humanoids
10:26 PM ET	@NobleMachines	Runway	@NobleMachines10:26 PM ETRunway @NobleMachines on X Tomorrow in San Francisco: Join our CTO and co-founder Wen-Loong Ma at the @runwayml AI Summit, on how world models are reshaping #robotics and #physicalAI. ➡️ The Masonic, 9am–6pm. Lineup and last-minute tickets: https://t.co/tp2N74ARbJ…	Physical AI
3:25 PM ET	@a16z	World Labs	@a16z3:25 PM ETWorld Labs @a16z on X World Labs co-founders Dr. Fei-Fei Li and Justin Johnson on compute as the constraint, and the Slack message that convinced them to go all-in on their Atlas model: Fei-Fei: "I think [we] have total conviction about the scaling law." "I do…	Physical AI
11:48 AM ET	@humanoidsdaily	Tesla	@humanoidsdaily11:48 AM ETTesla @humanoidsdaily on X We spoke with the founders of Feather, O-ID and Delta about what they’re building next. This week’s newsletter: their answers, Tesla’s Optimus progress and AMD’s World Labs deal. Get the weekly briefing in your inbox. Subscribe free →…	Humanoids
10:14 AM ET	@XRoboHub	AMD	@XRoboHub10:14 AM ETAMD @XRoboHub on X This is big… AMD is buying World Labs for $8.2B and bringing Dr. Fei-Fei Li on board. In her open letter, Li says she’ll become EVP and chief scientist once the deal closes, working directly with Lisa Su. This looks like a win-win: World…	Physical AI
8:03 AM ET	@stretchcloud	AMD	@stretchcloud8:03 AM ETAMD @stretchcloud on X $8.2 billion. That's what AMD is paying for a company that turns two years old this year and has shipped exactly one product. World Labs was Fei-Fei Li's bet that language models were solving the wrong problem, that understanding physical…	Physical AI
5:02 AM ET	@michellelsun		@michellelsun5:02 AM ET @michellelsun on X if you are building: - actuators - dexterous hands - training data for physical ai - self-driving labs - magnets - the deployment layer of robotics - reindustrialization - space manufacturing - energy and battery innovation -…	Physical AI
4:32 AM ET	@lukas_m_ziegler	World Labs	@lukas_m_ziegler4:32 AM ETWorld Labs @lukas_m_ziegler on X Massive week in robotics. I went through everything from World Labs, Skild AI, RealSense and even more. Get your coffee and see what happened in robotics space:	Physical AI
12:40 AM ET	@TheHumanoidHub	Figure	@TheHumanoidHub12:40 AM ETFigure @TheHumanoidHub on X Who deserves the valuation: Figure at $39B or World Labs at $8.2B? AMD is buying World Labs, Fei-Fei Li's spatial-intelligence startup, for $8.2B in all stock. Some context on how wild this is: - World Labs is barely 2 years old (founded…	Physical AI
10:30 PM ET	@Scobleizer		@Scobleizer10:30 PM ET @Scobleizer on X "CausalWM does not rush to generate the future. It first reasons out how objects move and how 3D geometric relations change, then continues to generate future frames from those intermediate results. It names this paradigm Causal…	Physical AI
7:12 PM ET	@a16z	World Labs	@a16z7:12 PM ETWorld Labs @a16z on X World Labs' Dr. Fei-Fei Li on why the 3D world and spatial intelligence are fundamentally different from language models: "Language is fundamentally a purely generated signal. There's no language out there. You don't go out in the nature…	Physical AI
5:03 PM ET	@MilkRoadAI	AMD	@MilkRoadAI5:03 PM ETAMD @MilkRoadAI on X AMD just made one of its biggest moves yet to challenge Nvidia. AMD acquired World Labs for roughly $8.2 billion, giving it access to Fei Fei Li’s team, technology, and research in world models. World Labs is building AI that can…	Physical AI
5:00 PM ET	@Biti8888	Tesla	@Biti88885:00 PM ETTesla @Biti8888 on X Physical AI in the wild Xinhua just put @XPENG_Global and @Tesla driving systems through Amsterdam: canals, trams, bikes, pedestrians. Both handled it smoothly. This is what Physical AI looks like outside the lab. It's no longer just…	Physical AI
4:51 PM ET	@lukas_m_ziegler	AMD	@lukas_m_ziegler4:51 PM ETAMD @lukas_m_ziegler on X 🚨 BREAKING: @AMD is buying @theworldlabs for $8.2 billion! 💰 Fei-Fei Li's spatial intelligence company, acquired by AMD in an all-stock deal, expected to close by the end of the year subject to regulatory approval. World Labs builds…	Physical AI
4:38 PM ET	@TheRundownAI	AMD	@TheRundownAI4:38 PM ETAMD @TheRundownAI on X ACQUIRED: AMD has agreed to buy World Labs, the 3D-world AI startup led by "godmother of AI" Fei-Fei Li, for $8.2B in stock. Li will become AMD's chief scientist, reporting to CEO Lisa Su. https://t.co/svhEDxesGZ	Physical AI
4:12 PM ET	@theworldlabs	World Labs	@theworldlabs4:12 PM ETWorld Labs @theworldlabs on X We are excited to announce that World Labs is joining @AMD. The research and technical breakthroughs we have achieved since our founding in 2024 have given us a clear vision for AI’s potential to solve problems in the spatial and physical…	Physical AI
10:35 AM ET	@PrismaXai		@PrismaXai10:35 AM ET @PrismaXai on X Day one at IROS, and booth 1122 is open! Teams training VLAs, world models, and robotics foundation models are stopping by to see our demo and talk through what they're building. #IROS2026 @ieeeiros https://t.co/6KmDgOpq5P	Physical AI
6:21 AM ET	@logi_today	Dynamic Map Platform	@logi_today6:21 AM ETDynamic Map Platform @logi_today on X DMP、3D地図をAI学習・評価の基盤に展開 https://t.co/MnzRgTLAXa 高精度3次元地図を、自動運転AIの「学習素材」に。DMPが走行シナリオの映像生成やデジタルツイン活用を紹介。天候や時間帯を変えた仮想環境で、AIの学習・評価・シミュレーションに生かす。	Physical AI
5:14 AM ET	@CyberRobooo	Unitree	@CyberRobooo5:14 AM ETUnitree @CyberRobooo on X GPT-6 Astra+Unitree G1->home-humanoid Stanford’s HomeBody is testing a different way to build humanoid intelligence. A Unitree G1 first explores an unfamiliar kitchen, builds persistent spatial memory and a Real2Sim digital twin, then uses…	Humanoids
9:27 PM ET	@AtMapshock	Momenta	@AtMapshock9:27 PM ETMomenta @AtMapshock on X The standard framing of Momenta as an "ADAS software company" understates what the firm is actually building. Momenta positions itself as a foundation model developer for physical AI. Its April 2026 R7 world model — deployed in mass…	Physical AI
3:55 PM ET	@michellelsun	Core Matter	@michellelsun3:55 PM ETCore Matter @michellelsun on X I’ll be at IROS this week and look forward to connecting with researchers and operators building dexterous hands, autonomous factories, world models. Core Matter @corematter_ is co-hosting Robotics Research Night with Saturday Robotics…	Humanoids
10:00 PM ET	@nguyenthambt	Vangrid	@nguyenthambt10:00 PM ETVangrid @nguyenthambt on X The most interesting part of Physical AI may not be the robots. It may be the infrastructure collecting the real-world data they need. @vangrid_io is approaching this through something almost everyone already has: a smartphone. Vangrid…	Physical AI
5:03 AM ET	@IlirAliu_		@IlirAliu_5:03 AM ET @IlirAliu_ on X What if sim and reality were one? This system keeps them in sync… always! 👉 Project & Paper ⬇️ > uses 3D Gaussians and particles to predict and correct object states > creates a real-time digital twin that stays synced with the real world…	Physical AI
9:20 AM ET	@chris_j_paxton		@chris_j_paxton9:20 AM ET @chris_j_paxton on X New open weight world action model-> https://t.co/vVkS2MPinM	Physical AI
5:49 AM ET	@Scobleizer		@Scobleizer5:49 AM ET @Scobleizer on X If you need video of people’s houses and other places for future world models that will run robots, this is what I would build. https://t.co/3yte7jTHTO	Physical AI
2:51 PM ET	@NVIDIARobotics	NVIDIA	@NVIDIARobotics2:51 PM ETNVIDIA @NVIDIARobotics on X Congrats to @bfl_ai on launching FLUX 3 Action and topping the RoboLab-120 leaderboard. 🙌 The fine-tunable world action model can learn policies for robot arms and a range of interactive environments. As a Cosmos Coalition member, BFL is…	Hands
2:49 PM ET	@NVIDIARobotics	NVIDIA	@NVIDIARobotics2:49 PM ETNVIDIA @NVIDIARobotics on X Congratulations to @bfl_ai on launching FLUX 3 Action and topping the RoboLab-120 leaderboard. 🙌 As a member of the Cosmos Coalition, BFL is working with NVIDIA to advance next-generation world models for physical AI. Try FLUX 3 Action on…	Physical AI
1:31 PM ET	@theworldlabs	World Labs	@theworldlabs1:31 PM ETWorld Labs @theworldlabs on X Block out a world. Let Atlas bring it to life. A sneak peek at Chisel, now in Atlas beta. https://t.co/ueaTdbItna	Physical AI