Skip to content
ContentLora

    Tip: press / anywhere to search.

    product

    NVIDIA Isaac GR00T

    Also known as GR00T, Project GR00T, GR00T N1, GR00T N2, Generalist Robot 00 Technology

    Isaac GR00T is NVIDIA's humanoid robot platform. It was announced in March 2024 as a general-purpose foundation model for humanoids,[1] and it now spans open GR00T models, the Isaac Sim and Isaac Lab simulators, and teleoperation tools.[2] As of October 2026, GR00T N1.7 is in early access and NVIDIA says GR00T N2 will ship by the end of the year.[3]

    Editor reviewedUpdated Robotics and embodied AIArtificial intelligence
    Key facts

    Isaac GR00T is NVIDIA‘s platform for humanoid robot developers. It combines robot foundation models, simulation tools and onboard compute, and many humanoid makers build on it.[1][4]

    Models

    NVIDIA announced Project GR00T on March 18, 2024 as “a general-purpose foundation model for humanoid robots”. It launched alongside the Jetson Thor onboard computer.[1] GR00T N1 (March 2025) was released as an open model. It pairs a vision-language module, which interprets the scene and instructions, with a diffusion-transformer module that generates motor actions in real time.[5] It was trained on real robot trajectories, human videos and synthetic data, and was demonstrated on the Fourier GR-1 humanoid.[6]

    At its GTC conference on March 16, 2026, NVIDIA made GR00T N1.7 available in early access with commercial licensing. It also previewed GR00T N2, which it says succeeds at new tasks in new environments more than twice as often as leading VLA models. N2 is slated for release by the end of 2026.[3]

    Between those releases came GR00T N1.6, announced on 5 January 2026 as an open reasoning VLA that adds full-body control and uses NVIDIA’s Cosmos Reason model for contextual understanding.[7] By October 2026 N1.7 was generally available under the Apache 2.0 licence, so companies can use it commercially. NVIDIA says it performs comparably to N1.6 but generalizes and follows language better, helped by 20,000 hours of human video in pretraining.[8] Its architecture still pairs a vision-language foundation model with a diffusion-transformer head that turns noise into continuous actions, and it is adapted to each robot through post-training.[9] N1.7 also describes actions as changes from the robot’s current hand pose rather than absolute targets, a format shared by robot and human data that NVIDIA credits for better cross-embodiment performance.[10]

    Simulation and data tools

    The GR00T platform bundles teleoperation tools for recording demonstrations, the Isaac Sim and Isaac Lab simulators, and robot middleware.[2] In March 2026 NVIDIA added Isaac Lab 3.0 and the Cosmos 3 world foundation model, which generates synthetic worlds and simulates actions.[11] This links GR00T to sim-to-real methods. The approach goes back to NVIDIA’s Isaac Gym (2021), which ran physics simulation and policy training together on the GPU and reported two to three orders of magnitude faster robot reinforcement learning than CPU-based simulators.[12]

    NVIDIA also works with Hugging Face. Isaac open models and libraries are integrated into Hugging Face’s LeRobot library, and its open-source Reachy 2 humanoid can run GR00T N1.6 on NVIDIA’s Jetson Thor computer.[13]

    Reference humanoid

    On May 31, 2026 NVIDIA announced the Isaac GR00T Reference Humanoid, an open design for academic research. It combines a Unitree H2 Plus body, Sharpa tactile hands and Jetson Thor compute, and Unitree is to sell it from late 2026.[14] Ai2, ETH Zurich, the Stanford Robotics Center and a UC San Diego lab plan to use it.[15]

    Partners

    NVIDIA lists 1X, AGIBOT, Agility, Boston Dynamics and Figure among humanoid developers building on its platform.[4]

    Questions readers ask

    Is GR00T open source?

    NVIDIA released GR00T N1 as an open foundation model for humanoid robots. GR00T N1.7 is offered in early access with commercial licensing.[5][3]

    What is GR00T N2?

    It is NVIDIA's next-generation robot model, previewed in March 2026. NVIDIA says it succeeds at new tasks in new environments more than twice as often as leading VLA models, and that it is slated for release by the end of 2026.[3]

    Does NVIDIA make a humanoid robot?

    NVIDIA designed a reference humanoid for researchers. It combines a Unitree H2 Plus body with Sharpa tactile hands and NVIDIA Jetson Thor compute, and Unitree will sell it from late 2026.[14]

    Sources

    Each numbered claim is a statement we checked against the sources listed with it. Status shows how well established it is.

    1. [1]

      NVIDIA announced Project GR00T, a general-purpose foundation model for humanoid robots, on March 18, 2024, alongside the Jetson Thor robot computer. confirmedas of 2024-03-18

    2. [2]

      NVIDIA's Isaac GR00T platform includes teleoperation tools for capturing demonstrations, open GR00T foundation models, and the Isaac Sim and Isaac Lab simulation and training tools. confirmedas of 2026-05-31

    3. [3]

      On March 16, 2026 NVIDIA made GR00T N1.7 available in early access with commercial licensing and previewed GR00T N2, which it says succeeds at new tasks in new environments more than twice as often as leading VLA models and is slated for release by the end of 2026. confirmedas of 2026-03-16

    4. [4]

      NVIDIA named 1X, AGIBOT, Agility, Boston Dynamics, Figure and others among humanoid developers building on its robotics platform in March 2026. confirmedas of 2026-03-16

    5. [5]

      GR00T N1 (March 2025) is an open humanoid foundation model with a vision-language module for understanding and a diffusion-transformer module that generates motor actions in real time. confirmedas of 2025-03-18

    6. [6]

      GR00T N1 was trained on a mixture of real-robot trajectories, human videos and synthetically generated data, and was deployed on the Fourier GR-1 humanoid. confirmedas of 2025-03-18

    7. [7]

      On 5 January 2026 NVIDIA released GR00T N1.6, an open reasoning VLA for humanoids that unlocks full-body control and uses NVIDIA's Cosmos Reason model. confirmedas of 2026-01-05

    8. [8]

      As of October 2026 GR00T N1.7 was generally available under the Apache 2.0 licence, with a new vision-language backbone and 20,000 hours of human video in pretraining; NVIDIA says it performs comparably to N1.6 with better generalization and language following. confirmedas of 2026-10-10

    9. [9]

      NVIDIA describes GR00T N1.7 as a vision-language foundation model combined with a diffusion-transformer head that denoises continuous actions, adaptable to specific robots through post-training. confirmedas of 2026-10-10

    10. [10]

      GR00T N1.7 represents actions as changes from the robot's current end-effector pose, a format shared across robot and human data that NVIDIA says improves cross-embodiment generalization. confirmedas of 2026-10-10

    11. [11]

      In March 2026 NVIDIA announced Cosmos 3, which it describes as a world foundation model combining synthetic world generation, vision reasoning and action simulation, and Isaac Lab 3.0 for large-scale robot learning. confirmedas of 2026-03-16

    12. [12]

      NVIDIA's Isaac Gym (2021) runs physics simulation and policy training on the GPU, which its authors report speeds up robot reinforcement learning by two to three orders of magnitude compared with CPU-based simulators. confirmedas of 2021-08-24

    13. [13]

      NVIDIA and Hugging Face integrated Isaac open models and libraries into Hugging Face's LeRobot library, and Hugging Face's Reachy 2 humanoid can run GR00T N1.6 on Jetson Thor. confirmedas of 2026-01-05

    14. [14]

      On May 31, 2026 NVIDIA announced the Isaac GR00T Reference Humanoid, an open research design that pairs a Unitree H2 Plus body with Sharpa tactile hands and Jetson Thor compute, to be available from Unitree in late 2026. confirmedas of 2026-05-31

    15. [15]

      Ai2, ETH Zurich, the Stanford Robotics Center and UC San Diego's Advanced Robotics and Controls Laboratory plan to use NVIDIA's reference humanoid. confirmedas of 2026-05-31

    16. [16]

      On July 30, 2026 Google DeepMind announced Gemini Robotics 2, a vision-language-action model it says can control full humanoids, including walking and manipulation, as well as bi-arm robots; previous models in the family controlled a humanoid's upper body for tabletop tasks. confirmedas of 2026-07-30

    17. [17]

      Physical Intelligence's π0 (October 2024) is a vision-language-action model that adds a flow-matching action generator on top of a pretrained vision-language model. confirmedas of 2024-10-31

    Revision history (2)
    1. Page created.
    2. Added GR00T N1.6 (January 2026), N1.7 general availability under Apache 2.0, the model architecture, Hugging Face LeRobot integration and the Isaac Gym lineage.

    Created Oct 10, 2026. Last reviewed by an editor on Oct 10, 2026. Next scheduled review: Jan 10, 2027.

    Cite this page

    "NVIDIA Isaac GR00T." ContentLora, updated Oct 10, 2026. https://contentlora.com/wiki/isaac-gr00t

    Spotted an error? Suggest a correction or emailcorrections@contentlora.com.

    Keep exploring

    Connections
    Unitree Robotics Reference humanoid body b…Figure AI Platform partnerGemini Robotics Competing robot modelPhysical Intelligen… Competing robot modelNVIDIA Developed by
    ☰ Contents