NVIDIA and Hugging Face Bring New Models and Frameworks to LeRobot for the Open Robotics Community
Summary
NVIDIA and Hugging Face are integrating NVIDIA's Isaac GR00T 1.7 robot foundation model, Isaac Teleop framework, and upcoming Cosmos 3 world model into the open-source LeRobot library to accelerate open robotics development.
<div id="bsf_rt_marker"></div><p><span style="font-weight: 400;">Open source AI has shown how quickly developers can innovate when models, data and tools are shared. </span><a target="_blank" href="https://www.nvidia.com/en-us/industries/robotics/"><span style="font-weight: 400;">Robotics</span></a><span style="font-weight: 400;"> has the same opportunity, but advancements in </span><a target="_blank" href="https://www.nvidia.com/en-us/glossary/generative-physical-ai/"><span style="font-weight: 400;">physical AI</span></a><span style="font-weight: 400;"> development can still be gated by costly and fragmented resources, from large datasets and robot foundation models to simulation, compute and validation tools.</span></p>
<p><span style="font-weight: 400;">NVIDIA and </span><a target="_blank" href="https://huggingface.co/blog/lerobot-release-v060"><span style="font-weight: 400;">Hugging Face</span></a><span style="font-weight: 400;"> are collaborating to bring the </span><a target="_blank" href="https://developer.nvidia.com/isaac/gr00t"><span style="font-weight: 400;">NVIDIA Isaac GR00T 1.7</span></a><span style="font-weight: 400;"> open, reasoning vision language action (VLA) model for </span><a target="_blank" href="https://www.nvidia.com/en-us/use-cases/humanoid-robots/"><span style="font-weight: 400;">humanoid robots</span></a><span style="font-weight: 400;"> and the </span><a target="_blank" href="https://nvidia.github.io/IsaacTeleop/"><span style="font-weight: 400;">NVIDIA Isaac Teleop</span></a><span style="font-weight: 400;"> framework to LeRobot — Hugging Face’s open source library for robotics — with </span><a target="_blank" href="https://www.nvidia.com/en-us/ai/cosmos/"><span style="font-weight: 400;">NVIDIA Cosmos 3</span></a><span style="font-weight: 400;">, a frontier model for physical AI, planned soon. Together, these integrations give developers a more accessible and standardized path for end-to-end robot development while driving innovation and collaboration across the open robotics community.</span></p>
<p><span style="font-weight: 400;">“Open source is how a field turns advanced research into something people can study, adapt and build on,” said Thomas Wolf, cofounder and chief science officer at Hugging Face. “With NVIDIA Isaac GR00T 1.7 and Isaac TeleOp in LeRobot today, robotics developers can use shared models, data and workflows to train and evaluate robots in the open. And with NVIDIA Cosmos 3 planned next, the community will have a path to bring frontier world models into that same collaborative loop.” </span></p>
<h2><b>An Open Pipeline for Robot Foundation Models</b></h2>
<p><span style="font-weight: 400;">Hugging Face LeRobot is an open source robotics library for training, running and sharing robot datasets, models, policies and workflows. NVIDIA’s continued partnership with Hugging Face connects NVIDIA’s 3 million robotics developers with Hugging Face’s 16 million AI builders, expanding access to frontier physical AI tools through open workflows.</span></p>
<p><span style="font-weight: 400;">Bringing NVIDIA physical AI capabilities into LeRobot gives developers a common way to collect and standardize data, train and fine-tune robot foundation models, evaluate performance and deploy models through open workflows. The integrations include:</span></p>
<ul>
<li style="font-weight: 400;" aria-level="1"><b>NVIDIA Isaac Teleop</b><span style="font-weight: 400;">, an open source framework for robot data collection, helps developers capture high-quality human demonstrations from external devices using standardized, interoperable formats, then expand and share datasets with the community, all directly in LeRobot. </span></li>
<li style="font-weight: 400;" aria-level="1"><b>NVIDIA Isaac GR00T 1.7</b><span style="font-weight: 400;">, the first open and commercially viable robot foundation model, makes it easier to post-train and deploy models through LeRobot workflows, helping developers adapt GR00T to new robot embodiments and tasks with benchmarked performance.</span></li>
<li style="font-weight: 400;" aria-level="1"><b>NVIDIA Cosmos 3</b><span style="font-weight: 400;">, a frontier world foundation model for physical AI coming soon to LeRobot, will help developers generate and augment robotics data, simulate scenarios and support policy development when real-world data is limited or too expensive to collect.</span></li>
</ul>
<p><span style="font-weight: 400;">These integrations build on a broader set of NVIDIA resources already connected to LeRobot to support the full robotics development loop, including: </span></p>
<ul>
<li style="font-weight: 400;" aria-level="1"><span style="font-weight: 400;">The largest </span><a target="_blank" href="https://huggingface.co/collections/nvidia/physical-ai"><b>open source physical AI dataset</b></a><span style="font-weight: 400;">, downloaded more than 15 million times, which includes more than 350,000 real and simulated trajectories and 57 million grasps to help developers kickstart their robotics workflows.</span></li>
<li style="font-weight: 400;" aria-level="1"><a target="_blank" href="https://developer.nvidia.com/isaac/sim"><b>NVIDIA Isaac Sim</b></a><b>– and </b><a target="_blank" href="https://developer.nvidia.com/isaac/lab"><b>Isaac Lab</b></a><b>-based simulation frameworks</b><span style="font-weight: 400;"> to help developers set up environments, generate robot data, test policies and validate behaviors before moving to physical robots.</span></li>
<li style="font-weight: 400;" aria-level="1"><a target="_blank" href="https://developer.nvidia.com/isaac/lab-arena"><b>NVIDIA Isaac Lab-Arena</b></a><b> in LeRobot Environment Hub</b><span style="font-weight: 400;"> to enable developers to quickly prototype complex simulation environments, register them in LeRobot EnvHub and seamlessly use them within the LeRobot ecosystem to train and evaluate generalist robot policies such as GR00T, Pi and SmolVLA.</span></li>
<li style="font-weight: 400;" aria-level="1"><b>NVIDIA Jetson Thor integration with LeRobot’s Reachy 2</b><span style="font-weight: 400;"> to support deployment of VLA models on open source humanoid robots.</span></li>
</ul>
<p><i><span style="font-weight: 400;">Learn more about how to use Isaac Teleop, Isaac GR00T 1.7 and Isaac Lab-Arena with LeRobot for end-to-end humanoid development and </span></i><a target="_blank" href="https://huggingface.co/blog/nvidia/nvidia-isaac-teleop-and-gr00t17-in-lerobot"><i><span style="font-weight: 400;">explore detailed LeRobot integration workflows</span></i></a><i><span style="font-weight: 400;">.</span></i></p>
View Cached Full Text
Cached at: 07/07/26, 10:37 AM
# NVIDIA and Hugging Face Bring New Models and Frameworks to LeRobot for the Open Robotics Community
Source: [https://blogs.nvidia.com/blog/hugging-face-lerobot-models-frameworks-open-robotics/](https://blogs.nvidia.com/blog/hugging-face-lerobot-models-frameworks-open-robotics/)
Open source AI has shown how quickly developers can innovate when models, data and tools are shared\.[Robotics](https://www.nvidia.com/en-us/industries/robotics/)has the same opportunity, but advancements in[physical AI](https://www.nvidia.com/en-us/glossary/generative-physical-ai/)development can still be gated by costly and fragmented resources, from large datasets and robot foundation models to simulation, compute and validation tools\.
NVIDIA and[Hugging Face](https://huggingface.co/blog/lerobot-release-v060)are collaborating to bring the[NVIDIA Isaac GR00T 1\.7](https://developer.nvidia.com/isaac/gr00t)open, reasoning vision language action \(VLA\) model for[humanoid robots](https://www.nvidia.com/en-us/use-cases/humanoid-robots/)and the[NVIDIA Isaac Teleop](https://nvidia.github.io/IsaacTeleop/)framework to LeRobot — Hugging Face’s open source library for robotics — with[NVIDIA Cosmos 3](https://www.nvidia.com/en-us/ai/cosmos/), a frontier model for physical AI, planned soon\. Together, these integrations give developers a more accessible and standardized path for end\-to\-end robot development while driving innovation and collaboration across the open robotics community\.
“Open source is how a field turns advanced research into something people can study, adapt and build on,” said Thomas Wolf, cofounder and chief science officer at Hugging Face\. “With NVIDIA Isaac GR00T 1\.7 and Isaac TeleOp in LeRobot today, robotics developers can use shared models, data and workflows to train and evaluate robots in the open\. And with NVIDIA Cosmos 3 planned next, the community will have a path to bring frontier world models into that same collaborative loop\.”
## **An Open Pipeline for Robot Foundation Models**
Hugging Face LeRobot is an open source robotics library for training, running and sharing robot datasets, models, policies and workflows\. NVIDIA’s continued partnership with Hugging Face connects NVIDIA’s 3 million robotics developers with Hugging Face’s 16 million AI builders, expanding access to frontier physical AI tools through open workflows\.
Bringing NVIDIA physical AI capabilities into LeRobot gives developers a common way to collect and standardize data, train and fine\-tune robot foundation models, evaluate performance and deploy models through open workflows\. The integrations include:
- **NVIDIA Isaac Teleop**, an open source framework for robot data collection, helps developers capture high\-quality human demonstrations from external devices using standardized, interoperable formats, then expand and share datasets with the community, all directly in LeRobot\.
- **NVIDIA Isaac GR00T 1\.7**, the first open and commercially viable robot foundation model, makes it easier to post\-train and deploy models through LeRobot workflows, helping developers adapt GR00T to new robot embodiments and tasks with benchmarked performance\.
- **NVIDIA Cosmos 3**, a frontier world foundation model for physical AI coming soon to LeRobot, will help developers generate and augment robotics data, simulate scenarios and support policy development when real\-world data is limited or too expensive to collect\.
These integrations build on a broader set of NVIDIA resources already connected to LeRobot to support the full robotics development loop, including:
- The largest[**open source physical AI dataset**](https://huggingface.co/collections/nvidia/physical-ai), downloaded more than 15 million times, which includes more than 350,000 real and simulated trajectories and 57 million grasps to help developers kickstart their robotics workflows\.
- [**NVIDIA Isaac Sim**](https://developer.nvidia.com/isaac/sim)**– and**[**Isaac Lab**](https://developer.nvidia.com/isaac/lab)**\-based simulation frameworks**to help developers set up environments, generate robot data, test policies and validate behaviors before moving to physical robots\.
- [**NVIDIA Isaac Lab\-Arena**](https://developer.nvidia.com/isaac/lab-arena)**in LeRobot Environment Hub**to enable developers to quickly prototype complex simulation environments, register them in LeRobot EnvHub and seamlessly use them within the LeRobot ecosystem to train and evaluate generalist robot policies such as GR00T, Pi and SmolVLA\.
- **NVIDIA Jetson Thor integration with LeRobot’s Reachy 2**to support deployment of VLA models on open source humanoid robots\.
*Learn more about how to use Isaac Teleop, Isaac GR00T 1\.7 and Isaac Lab\-Arena with LeRobot for end\-to\-end humanoid development and*[*explore detailed LeRobot integration workflows*](https://huggingface.co/blog/nvidia/nvidia-isaac-teleop-and-gr00t17-in-lerobot)*\.*
Similar Articles
@NVIDIARobotics: Robotics is built on collaboration. Hear @huggingface CSO & Co-founder Thomas Wolf explain how open source gives develo…
LeRobot v0.6.0 introduces new world models, reward models, faster data loading, and improved benchmarks for robot learning, closing the robot learning loop.
@seclink: 5. Open-Source Acceleration of Robot World Models - NVIDIA Cosmos 3 + Isaac GR00T: Physical AI Foundation Models - AGIBOT Genie Sim 3.0: The First Fully Open-Source Robot Simulation Platform (Complete Open Source of Code, Data, and Assets) - VLA (Vision-…
Robot world models and simulation platforms are experiencing open-source acceleration: NVIDIA launched Cosmos 3 and Isaac GR00T physical AI foundation models, AGIBOT released Genie Sim 3.0, a fully open-source simulation platform, VLA models become mainstream for manipulation policies, collectively lowering the entry barrier for the robotics field.
@lukas_m_ziegler: Another win for open-source robotics! @huggingface just released a fully open-source humanoid robot, and you can build …
Hugging Face released LeRobot Humanoid, an open-source, low-cost ($2,500) 3D-printed humanoid robot platform, including hardware, simulation, and control tools for robot learning research.
LeRobot v0.6.0: Imagine, Evaluate, Improve
LeRobot v0.6.0 is a major release of Hugging Face's robot learning library, adding world model policies (VLA-JEPA, FastWAM, LingBot-VA), new VLAs, reward models, six simulation benchmarks, depth sensing, VLM-powered dataset annotation, custom video encoding, cloud training, and a deployment CLI for human-in-the-loop corrections.
LeRobot v0.5.0: Scaling Every Dimension
LeRobot v0.5.0 is a major release featuring support for Unitree G1 humanoid robots, new policy architectures (Pi0-FAST VLAs, Real-Time Chunking), streaming video encoding for 3x faster training, and EnvHub for loading simulation environments from Hugging Face Hub.