NVIDIA has released new open-source physical AI models and the Jetson T4000 compute module at CES 2026, enabling developers to build smarter, more adaptable robots. Global partners including Boston Dynamics, Caterpillar, LG Electronics, and NEURA Robotics unveiled robots powered by NVIDIA’s robotics stack, signaling what CEO Jensen Huang calls “the ChatGPT moment for robotics”.
What NVIDIA Announced
NVIDIA released multiple open models for robot learning, reasoning, and simulation, all available on Hugging Face. The Cosmos Transfer 2.5 and Cosmos Predict 2.5 models enable physically based synthetic data generation and robot policy evaluation in simulation. Cosmos Reason 2, a reasoning vision language model, allows robots to see, understand, and act in the physical world like humans.
The Isaac GR00T N1.6 model is purpose-built for humanoid robots, providing full-body control and contextual understanding. Companies like Franka Robotics, NEURA Robotics, and Humanoid are using GR00T-enabled workflows to simulate, train, and validate new robot behaviors.
NVIDIA also launched Isaac Lab-Arena, an open-source framework for large-scale robot policy evaluation and benchmarking in simulation. The OSMO cloud-native orchestration framework unifies robotic development into a single command center, letting developers run workflows across different compute environments.
Jetson T4000 Brings Blackwell to Robotics
The Jetson T4000 module delivers 4x the energy efficiency of the previous generation with 1,200 FP4 TFLOPS and 64GB of memory at $1,999 for 1,000-unit volumes. Built on the NVIDIA Blackwell architecture, it offers a cost-effective upgrade path for Jetson Orin customers.
Humanoid robot developers showcased new robots integrated with Jetson Thor at CES. NEURA Robotics launched a Porsche-designed Gen 3 humanoid, Richtech Robotics unveiled Dex for industrial environments, and LG Electronics debuted a home robot for household tasks.
Boston Dynamics, Humanoid, and RLWRLD integrated Jetson Thor into existing humanoids to enhance navigation and manipulation. NVIDIA IGX Thor, available later in January 2026, extends robotics to the industrial edge with enterprise software support and functional safety.
Why This Matters for Robotics
Physical AI breakthroughs are unlocking new applications by enabling robots to understand the real world, reason, and plan actions. NVIDIA’s open models allow developers to bypass resource-intensive pretraining and focus on creating next-generation AI robots.
The collaboration with Hugging Face integrates Isaac and GR00T technologies into the LeRobot open-source framework, connecting NVIDIA’s 2 million robotics developers with Hugging Face’s 13 million AI builders. Robotics is now the fastest-growing category on Hugging Face, with NVIDIA’s open models leading downloads.
Real-world applications are already emerging. Salesforce uses Cosmos Reason with Agentforce to analyze video footage from its robots, reducing incident resolution times by 2x. LEM Surgical uses Cosmos Transfer to train the autonomous arms of its Dynamis surgical robot powered by Jetson AGX Thor.
Industry Partners and Use Cases
| Company | Robot/Application | NVIDIA Technology |
|---|---|---|
| Boston Dynamics | Enhanced existing humanoids | Jetson Thor |
| LG Electronics | Home robot for household tasks | Jetson Thor |
| NEURA Robotics | Porsche-designed Gen 3 humanoid | Jetson Thor, GR00T |
| Richtech Robotics | Dex mobile humanoid | Jetson Thor |
| Caterpillar | Construction and mining equipment AI | NVIDIA robotics stack |
| LEM Surgical | Dynamis surgical robot | Jetson AGX Thor, Cosmos Transfer |
Caterpillar is expanding its collaboration with NVIDIA to bring advanced AI and autonomy to construction and mining equipment and job sites. Archer is using IGX Thor to advance aircraft safety, airspace integration, and autonomy-ready systems in aviation.
What Developers Get Access To
NVIDIA released new open-source frameworks on GitHub that simplify complex pipelines and accelerate the transition from research to real-world use cases. Isaac Lab-Arena connects to industry-leading benchmarks like Libero and Robocasa, standardizing testing before deployment to physical hardware.
OSMO lets developers define and run workflows such as synthetic data generation, model training, and software-in-the-loop testing across different compute environments. Robot developers like Hexagon Robotics now use OSMO, which is integrated into the Microsoft Azure Robotics Accelerator toolchain.
Hugging Face’s open-source Reachy 2 humanoid will be fully interoperable with the NVIDIA Jetson Thor robotics computer, letting developers run any vision-language-action model, including GR00T N1.6. GR00T N models and Isaac Lab-Arena are now available in the LeRobot library for easy fine-tuning and evaluation.
What’s Next
Partners including AAEON, Advantech, ADLINK, Aetina, and others offer Thor-powered systems equipped for edge AI, robotics, and embedded applications. NVIDIA IGX Thor will be available later in January 2026 for industrial edge deployments.
Caterpillar CEO Joe Creed will share details about the expanded NVIDIA collaboration during a CES keynote on Wednesday, January 7, alongside NVIDIA Vice President of Robotics and Edge AI Deepu Talla. More announcements about practical applications in construction, mining, and industrial automation are expected.
The full stack of Jetson robotics processors, CUDA, Omniverse, and open physical AI models positions NVIDIA’s global ecosystem to transform industries with AI-driven robotics. All new models are available on Hugging Face with permissive open-weight licenses.
Featured Snippet Boxes
What is NVIDIA Cosmos for robotics?
NVIDIA Cosmos is a platform of generative world foundation models that enable synthetic data generation and robot policy evaluation in simulation. It includes Transfer, Predict, and Reason models that help robots understand the physical world and learn tasks faster without expensive real-world training.
How much does the NVIDIA Jetson T4000 cost?
The Jetson T4000 module is priced at $1,999 for 1,000-unit volumes. It delivers 1,200 FP4 TFLOPS, 64GB of memory, and 4x greater energy efficiency than the previous generation, making it cost-effective for autonomous machines and general robotics applications.
What is Isaac GR00T N1.6 used for?
Isaac GR00T N1.6 is an open reasoning vision-language-action model purpose-built for humanoid robots. It enables full-body control and uses Cosmos Reason for better reasoning and contextual understanding, allowing humanoids to learn coordination, dexterity, and navigation skills faster.
Which companies are building robots with NVIDIA technology?
Boston Dynamics, LG Electronics, NEURA Robotics, Richtech Robotics, Franka Robotics, Humanoid, Caterpillar, and LEM Surgical are using NVIDIA’s robotics stack. These partners are deploying Jetson Thor, GR00T models, and Isaac frameworks for humanoids, surgical robots, industrial equipment, and autonomous machines.
