# What is true on World Models?

> Markdown twin of https://roboterminal.app/topics/world-models · https://roboterminal.app/topics/world-models.md

## On this layer

- org — World Labs (12)
- org — a16z (5)
- org — [NVIDIA](https://roboterminal.app/companies/nvidia) (5)
- org — 2077 Ventures (2)
- org — Khosla Ventures (2)
- org — Promise (2)
- org — Radical Ventures (2)
- org — Veeda AI (2)
- org — a16z New Media (1)
- org — ADLINK (1)
- org — Anthropic (1)
- org — Atlas (1)
- org — Atlas model for spatial intelligence. (1)
- org — Axis Robotics (1)
- org — Booster Robotics (1)
- org — Columbia University (1)
- org — development timeline of Moby. (1)
- org — development timeline reduction. (1)
- org — DROID dataset robot interaction data (1)
- org — filtered simulation episodes (1)
- org — G1 humanoid grasp and lift task (1)
- org — [Google](https://roboterminal.app/companies/google) (1)
- org — Harvard University (1)
- org — historic crime scenes (1)
- org — Hydra-0 training (1)
- org — investment in Veeda (1)
- org — LimX Dynamics (1)
- org — Moby deployment in a semiconductor facility. (1)
- org — Noble Machines (1)
- org — OpenAI (1)
- org — Pixomondo (1)
- org — robot interaction data from the DROID dataset (1)
- org — Schaeffler (1)
- org — Seeed (1)
- org — Solomon (1)
- org — successful grasps by the G1 humanoid on hardware. (1)
- org — TRON 2 (1)
- org — Veeda (1)
- org — video generation (1)
- org — ZiNovaLabs (1)

## Who captures

- World Labs (8)
- NVIDIA (4)
- Joseph Amigo, Rooholla Khorrambakht, Nicolas Mansard, Ludovic Righetti (3)
- a16z (2)
- Veeda AI (2)
- Promise (2)
- @chris_j_paxton (1)
- @RoboPapers (1)
- @Scobleizer (1)
- @rimtoln (1)
- Anthropic (1)
- ZiNovaLabs (1)
- @drfeifei (1)
- Axis Robotics (1)
- @EdmondIsARobot (1)
- Edmond (1)

## Sourced numbers

- **Hydra-0 training:** 2202 hrs
  - _"It was trained on about 2,202 hours of multi-embodiment video spanning human hands, handheld grippers, single-arm robots and bimanual systems."_
  - Source: [@techniahqrobot on X](https://x.com/techniahqrobot/status/2095145553482891753)
- **TRON 2:** 1 units
  - _"TRON 2 serves as a modular and extensible embodied robotic platform for multi-tool, multi-step tasks across large workspaces."_
  - Source: [LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows.

@ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, includi… / X](https://x.com/chris_j_paxton/status/2095114381818253768)
- **Moby deployment in a semiconductor facility.:** 3 units
  - _"bring two Moby3 units to a semiconductor facility for material-handling workflows."_
  - Source: [Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA](https://x.com/NVIDIARobotics/status/2094923298589024698)
- **development timeline of Moby.:** $3
  - _"Accelerate Moby's development timeline by nearly 3x, from an expected four years to 18 months."_
  - Source: [Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA](https://x.com/NVIDIARobotics/status/2094923298589024698)
- **development timeline reduction.:** 18 hrs
  - _"from an initial estimated timeline of four years with 50 people to 18 months with 15 people."_
  - Source: [Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA](https://x.com/NVIDIARobotics/status/2094923298589024698)
- **Atlas model for spatial intelligence.:** 1 units
  - _"Atlas is our scalable foundation model that can perceive, generate, reason, and interact with virtual and physical worlds."_
  - Source: [World Labs on X: "All of this is powered by one unified architecture: a multimodal autoregressive diffusion transformer, pretrained from scratch.

This foundation blends the best of modern LLMs and video models, benefiting from the architectural, algorithmic, and systems advances from both areas." / X](https://x.com/theworldlabs/status/2094839768115036581)
- **Atlas:** 1 units
  - _"Atlas is built to scale: its performance improves with increased training compute, and we expect this trend to hold as we continue scaling."_
  - Source: [Atlas: A World Model for Spatial Intelligence | World Labs](https://x.com/theworldlabs/status/2094839769616589051)
- **@mushman1970 2094416619275473057:** $90M
  - _"$90M"_
  - Source: [Veeda AI Raises $90M Seed to Build "The Matrix" for Physical AI - WOWTALE](https://x.com/mushman1970/status/2094416619275473057)
- **investment in Veeda:** $90million
  - _"Sanja Fidler quit her job running NVIDIA's Toronto AI lab, started a company three months ago, and has already pulled in $90 million without shipping a single product."_
  - Source: [Shuddhatma Jain on X: "Sanja Fidler quit her job running NVIDIA's Toronto AI lab, started a company three months ago, and has already pulled in $90 million without shipping a single product.

nobody is betting on a robot here. they're betting she can build the world robots learn inside 🧵" / X](https://x.com/decepticons0101/status/2092904684730081387)
- **filtered simulation episodes:** 180.55 hrs
  - _"42,046 episodes and 3,249,835 frames remained, equivalent to 180.55 hours at 5 Hz."_
  - Source: [Axis Robotics × Booster: From Digital Twins to a Robot Data Engine](https://x.com/axisrobotics/status/2092884880430661971)
- **robot interaction data from the DROID dataset:** 62 hrs
  - _"Meta reports using 62 hours of robot interaction data from the DROID dataset for this stage."_
  - Source: [🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma](https://x.com/EdmondIsARobot/status/2092037991619530779)
- **historic crime scenes:** 10 units
  - _"to recreate ten historic crime scenes that no longer existed."_
  - Source: [From Archive to Production: A Hybrid Workflow with Marble | World Labs](https://x.com/theworldlabs/status/2090479434927411551)
- **G1 humanoid grasp and lift task:** 10 units
  - _"G1 humanoid · grasp & lift · ten successes in a row"_
  - Source: [Coupled Local and Global World Models for Efficient First Order RL](https://coupled-global-local-wm-rl.pages.dev/)
- **successful grasps by the G1 humanoid on hardware.:** 10 units
  - _"G1 humanoid · grasp & lift · ten successes in a row"_
  - Source: [Coupled Local and Global World Models for Efficient First Order RL](https://x.com/RoboPapers/status/2092642979466915921)
- **DROID dataset robot interaction data:** 62 hrs
  - _"Meta reports using 62 hours of robot interaction data from the DROID dataset for this stage."_
  - Source: [🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma](https://luma.com/5h9owhri)
- **video generation:** 1 units
  - _"Atlas generates images and videos from one or more images with pixel-perfect camera control, outputting up to 1 minute of video at 1440p."_
  - Source: [World Labs Atlas](https://www.worldlabs.ai/blog/atlas)

## Pulse

- [@a16z on X](https://x.com/a16z/status/2095972688200372368)
  >blogs about ray tracing for a CS project in 2012 >co-founds World Labs in 2024 >ships Atlas in 2026, a spatial intelligence model that turns a few photos into a whole 3D world curiosity comp…
- [@a16z on X](https://x.com/a16z/status/2095961376208646248)
  >be dr. fei-fei li
>born in beijing, raised in chengdu
>dad moves to new jersey, follow at age 16
>speak almost no english
>parents open a dry cleaner
>work the counter on weekends
>get into princeton…
- [@a16z on X](https://x.com/a16z/status/2095940012932215128)
  World Labs co-founder Justin Johnson says Atlas can recreate the famous "Bullet Time" shot from The Matrix with iPhones: "The way they did that shot is they had a ring of hundreds of cameras. [Neo] fe…
- [@drfeifei on X](https://x.com/drfeifei/status/2095926761305575826)
  Next view prediction is the key to Atlas, enabling us to unify pixel-level generation and reconstruction. @jcjohnss @BenMildenhall @martin_casado and I had a deeper discussion on some of the most exci…
- [@a16z on X](https://x.com/a16z/status/2095921217308086425)
  Old footage sitting in your camera roll could now be reconstructable as a 3D scene. World Labs co-founders Ben Mildenhall and Fei-Fei Li on how Atlas got there: Ben: "In a casual sense... I took three…
- [@a16z on X](https://x.com/a16z/status/2095882301032828932)
  World Labs co-founders Fei-Fei Li, Justin Johnson, Ben Mildenhall, and a16z's Martin Casado on Atlas, a world model for spatial intelligence: LLMs are built on next token prediction. Video models are …
- [@chris_j_paxton on X](https://x.com/chris_j_paxton/status/2095854947938058250)
  Learn contact-rich humanoid manipulation policies, purely within a world model. This is usually not tractable, because you need a very large model to get the contact dynamics right -- instead, @Jspham…
- [@RoboPapers on X](https://x.com/RoboPapers/status/2095854394667393218)
  Reinforcement learning has unlocked tremendous gains in broader AI and humanoid robot whole body control, but still has yet to deliver strong results for robot manipulation due to the difficulty of ca…
- [@Scobleizer on X](https://x.com/Scobleizer/status/2095756331915542758)
  A 3D world for your bots. https://t.co/WSUa10DjhA
- [@rimtoln on X](https://x.com/rimtoln/status/2095568388684738657)
  EVERYONE SHOWS THE 300-AGENT SWARM. almost nobody shows what happens after 300 agents are easy to spawn turning 300 outputs into one structure is the hard part ▹ the pipeline nobody demos collect → co…
- [@MParekh on X](https://x.com/MParekh/status/2095258950320996639)
  ‘New and Improved’ in AI Models, Anthropic, OpenAI & Google. ARD #154 ...Fable 5.1’s fine print. Astra tips out on cyber. Gemini narrows the gap. World models level up. Full writeup and source lin…
- [@techniahqrobot on X](https://x.com/techniahqrobot/status/2095145553482891753)
  We’re cooked if robots can start learning from bodies they don’t even have. NVIDIA Brown Columbia and Harvard researchers released Hydra-0, a world model that represents robot actions as pixel motion.…
- [LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows.

@ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, includi… / X](https://x.com/chris_j_paxton/status/2095114381818253768)
  Post Log in Sign up Post LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows. @ZiNovaLabs builds on TRON 2 to explore an innovati…
- [Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA](https://x.com/NVIDIARobotics/status/2094923298589024698)
  Manufacturing \| Robotics Noble Machines Accelerates General Purpose Industrial Robot Development With NVIDIA Isaac GR00T Learn More Objective Noble Machines is advancing industrial robot development …
- [@drfeifei on X](https://x.com/drfeifei/status/2094910083444707551)
  @DrJimFan @theworldlabs Indeed Jim, a camera conditioned world model with spatial context has so much horizontal usage including real2sim for robotics!
- [Seeed reBot Arm B601-RS: Physical AI & VLA Model Course | NVIDIA DLI Series](https://x.com/NVIDIARobotics/status/2094850247184826558)
  Warehouse China Warehouse US Warehouse Germany Warehouse The store will not work correctly in the case when cookies are disabled. US Warehouse: Enjoy FREE UNIUNI shipping on orders under $50! (Exclude…
- [World Labs on X: "All of this is powered by one unified architecture: a multimodal autoregressive diffusion transformer, pretrained from scratch.

This foundation blends the best of modern LLMs and video models, benefiting from the architectural, algorithmic, and systems advances from both areas." / X](https://x.com/theworldlabs/status/2094839768115036581)
  Post Log in Sign up Post World Labs on X: "All of this is powered by one unified architecture: a multimodal autoregressive diffusion transformer, pretrained from scratch. This foundation blends the be…
- [Atlas: A World Model for Spatial Intelligence | World Labs](https://x.com/theworldlabs/status/2094839769616589051)
  September 1, 2026Introducing Atlas, our new omni world model for spatial intelligence. Atlas: A World Model for Spatial Intelligence World models generate, reconstruct, and simulate any possible world…
- [Veeda AI Raises $90M Seed to Build "The Matrix" for Physical AI - WOWTALE](https://en.wowtale.net/2026/08/31/234943)
  Skip to content - AI - · - Funding - · - Robotics Veeda AI Raises $90M Seed to Build “The Matrix” for Physical AI 8/31/2026 ・ 버섯돌이(Taehyun Kim) Training a robot to master a new motion takes thousands …
- [Veeda AI Raises $90M Seed to Build "The Matrix" for Physical AI - WOWTALE](https://x.com/mushman1970/status/2094416619275473057)
  Skip to content - AI - · - Funding - · - Robotics Veeda AI Raises $90M Seed to Build “The Matrix” for Physical AI 8/31/2026 ・ 버섯돌이(Taehyun Kim) Training a robot to master a new motion takes thousands …
- [Shuddhatma Jain on X: "Sanja Fidler quit her job running NVIDIA's Toronto AI lab, started a company three months ago, and has already pulled in $90 million without shipping a single product.

nobody is betting on a robot here. they're betting she can build the world robots learn inside 🧵" / X](https://x.com/decepticons0101/status/2092904684730081387)
  Post Log in Sign up Post Shuddhatma Jain on X: "Sanja Fidler quit her job running NVIDIA's Toronto AI lab, started a company three months ago, and has already pulled in $90 million without shipping a …
- [Axis Robotics × Booster: From Digital Twins to a Robot Data Engine](https://x.com/axisrobotics/status/2092884880430661971)
  Back How task-aligned simulation, real demonstrations, and continued pretraining can make each new Booster task easier to build than the last. Robot learning has a model problem. More importantly, it …
- [🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma](https://x.com/EdmondIsARobot/status/2092037991619530779)
  Hosted By Edmond Jono Hart James (Jingxi) Xu Mene Mazarakis Angela Zhang Manfredi Bernardi Adam 134 Went Akshobhya Gupta, Hannah Tsui and 132 others Contact the Host Report Event AI 🤖🍨 Sundae Roboti…
- [Bringing Marble to Life | World Labs](https://x.com/theworldlabs/status/2090479436550602863)
  Nov 12, 2025A behind-the-scenes look at how Marble powered the creation of its own launch story. Bringing Marble to Life World Labs: Just Imagine - YouTube Tap to unmute World Labs: Just Imagine World…
- [From Archive to Production: A Hybrid Workflow with Marble | World Labs](https://x.com/theworldlabs/status/2090479434927411551)
  Aug 12, 2026How Promise used Marble to transform archival imagery into immersive 3D environments for Harlan Coben’s Final Twist. From Archive to Production: A Hybrid Workflow with Marble Using Hybrid …
- [Coupled Local and Global World Models for Efficient First Order RL](https://coupled-global-local-wm-rl.pages.dev/)
  Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and E…
- [Coupled Local and Global World Models for Efficient First Order RL](https://x.com/RoboPapers/status/2092642979466915921)
  Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and E…
- [Coupled Local and Global World Models for Efficient First Order RL](https://x.com/RoboPapers/status/2093005367567347970)
  Real-world tasks solved zero-shot by policies trained entirely inside our learned world models: Push-T with a tabletop manipulator (left), Ego-Centric Grasp and Lift with a G1 humanoid (centre), and E…
- [From Archive to Production: A Hybrid Workflow with Marble | World Labs](https://www.worldlabs.ai/case-studies/from-archive-to-production)
  Aug 12, 2026How Promise used Marble to transform archival imagery into immersive 3D environments for Harlan Coben’s Final Twist. From Archive to Production: A Hybrid Workflow with Marble Using Hybrid …
- [🤖🍨 Sundae Robotics 06: V-JEPA 2 & Predicting Physical Intelligence · Luma](https://luma.com/5h9owhri)
  Hosted By Edmond Jono Hart James (Jingxi) Xu Mene Mazarakis Angela Zhang Manfredi Bernardi Adam 134 Went Akshobhya Gupta, Hannah Tsui and 132 others Contact the Host Report Event AI 🤖🍨 Sundae Roboti…
- [World Labs Atlas](https://www.worldlabs.ai/blog/atlas)
  September 1, 2026Introducing Atlas, our new omni world model for spatial intelligence. Atlas: A World Model for Spatial Intelligence World models generate, reconstruct, and simulate any possible world…

---

Source: Roboterminal · https://roboterminal.app/topics/world-models.md