Identity
- Status
- active
- Developer ID
nvidia/GR00T-N1.7-3B- Released
- Jul 7, 2026
- Version
- N1.7
- License
- apache-2.0
NVIDIAGR00T N1.7 3B
NVIDIA's open vision-language-action base model for generalized humanoid skills and post-training on new tasks.
- Model type
- Vision-language-action model
- Inputs
- Text, Image, Robot state
- Inference
- On device
- Released
- Jul 7, 2026
Providers & Pricing
0 current hosted offerings
No Hosted API Available
This record is downloadable or self-hosted. Its cost depends on the infrastructure and serving configuration, not a universal model price.
At a Glance
Model Facts
Model Design
- Model type
- Vision-language-action model
- Inputs
- Text, Image, Robot state
- Outputs
- Robot action
- Total parameters
- 3B
- Architecture
- Vision-language backbone with action expert
Deployment
- Inference
- On device
- Control rate
- Unknown
- API available
- No
- Open weights
- Yes
- Self-hostable
- Yes
Publisher Media
GR00T N1.7 3B Demonstrations
Embodied Intelligence
Robotics Control Profile
Control Contract
Model typeVision-language-action model
Action representationPredictive chunks of relative joint motions
Control architectureVision-language backbone with action expert
InferenceOn device
Native control rateUnknown
Training Data and Embodiments
Mixture of real teleoperation, synthetic robot data, and internet-scale video described by NVIDIA.
Verified Platform Links
Compatible Robots
| Robot | Relationship | Form | Availability | Evidence note | Source |
|---|---|---|---|---|---|
 G1Unitree Robotics | Training embodiment | Humanoid | Commercially available | NVIDIA lists Unitree G1 among the embodiments represented in the GR00T training and adaptation ecosystem. | Official fact |
This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Version Lineage
PredecessorUnknown
SuccessorsUnknown
Aliasesnvidia/GR00T-N1.7-3B, nvidia/GR00T-N1.7-3B, nvidia/GR00T-N1.7-3B
Explicit Unknowns
- Native control rate
Primary Evidence
Sources and Observation Date
NVIDIA Isaac GR00T N1.7 official repositoryNVIDIA · observed Aug 29, 2026nvidia/GR00T-N1.7-3B config.json at 2fc962b973bcNVIDIA · observed Aug 28, 2026nvidia/GR00T-N1.7-3B developer model repositoryNVIDIA · observed Aug 28, 2026nvidia/GR00T-N1.7-3B README.md at 2fc962b973bcNVIDIA · observed Aug 28, 2026nvidia/GR00T-N1.7-3B immutable Hub API metadataNVIDIA · observed Aug 28, 2026Hugging Face catalogHugging Face · observed Sep 2, 2026
Comparable Peers
Related Model Comparisons
| A | Pair | B | Context |
|---|---|---|---|
GR00T N1.7 3BNVIDIA | vs |  MolmoAct 2Ai2 | cross-developer peersrobotics |
GR00T N1.7 3BNVIDIA | vs |  Helix 02Figure | cross-developer peersrobotics |
GR00T N1.7 3BNVIDIA | vs |  Gemini Robotics 2Google DeepMind | cross-developer peersrobotics |
GR00T N1.7 3BNVIDIA | vs |  SmolVLA 450MHugging Face LeRobot | cross-developer peersrobotics |
GR00T N1.7 3BNVIDIA | vs |  OpenVLA 7BOpenVLA Research Team | cross-developer peersrobotics |
GR00T N1.7 3BNVIDIA | vs |  pi 0.7Physical Intelligence | cross-developer peersrobotics |
Questions
GR00T N1.7 3B FAQs
What is GR00T N1.7 3B?+
NVIDIA's open vision-language-action base model for generalized humanoid skills and post-training on new tasks. It is developed by NVIDIA and its current sourced lifecycle status is active.
What inputs and outputs does GR00T N1.7 3B support?+
GR00T N1.7 3B's current record lists Text, Image, and Robot state as input and Robot action as output.
What kind of robotics model is GR00T N1.7 3B?+
GR00T N1.7 3B is classified as a vision-language-action model. Its action representation is Predictive chunks of relative joint motions, its control architecture is Vision-language backbone with action expert, and inference is recorded as on device.
Which robots and embodiments work with GR00T N1.7 3B?+
The developer reports Unitree G1, AgiBot Genie-1, Fourier GR-1, and bimanual manipulation platforms as supported or represented embodiments. Model Markets also has sourced model-to-platform links for G1; each link states whether it is native, demonstrated, used for training, or used for evaluation.
How large is GR00T N1.7 3B?+
The current record lists 3B total parameters and an unknown active parameter count. Its architecture is Vision-language backbone with action expert.
What capabilities and tasks does GR00T N1.7 3B support?+
GR00T N1.7 3B's recorded capabilities are cross-embodiment, dexterous-manipulation, whole-body-control, and fine-tuning. Its supported tasks are robot-control, dexterous-manipulation, and whole-body-control.
What training data is reported for GR00T N1.7 3B?+
Mixture of real teleoperation, synthetic robot data, and internet-scale video described by NVIDIA.
Is GR00T N1.7 3B available through an API?+
GR00T N1.7 3B is explicitly marked as not API-available in the current record.
How much does GR00T N1.7 3B cost through an API?+
GR00T N1.7 3B has no verified hosted API offering, so no provider token price applies. It is a downloadable or self-hosted artifact whose cost depends on the serving infrastructure and workload.
Can GR00T N1.7 3B be self-hosted?+
GR00T N1.7 3B is an open-weight model. Self-hosting is supported, and the recorded license is apache-2.0.