Isaac GR00T N1.7
Open 3B cross-embodiment VLA for humanoids and manipulators.
GR00T N1.7 (early access) is NVIDIA’s open vision-language-action model. A Cosmos-Reason2 VLM backbone reads camera frames and a language instruction; a flow-matching diffusion-transformer action head outputs continuous motor commands for the robot’s embodiment. Post-train it on real or synthetic data for a specific robot.
- Provider
- NVIDIA
- Type
- Robotics (VLA)
- Released
- Apr 17, 2026
- Price
- Open weights — self-host
- Input
- text, image
- Output
- action
- Open weights
- Yes
- License
- NVIDIA Open Model License
Best for
- Agents
- Vision
- Local & on-device
Strengths
- Open weights, commercial use allowed
- Cross-embodiment
- 3B parameters
More from NVIDIA
- Nemotron 3 Super — NVIDIA. Open hybrid Mamba-Transformer MoE for multi-agent applications.
- Nemotron 3.5 Lightning — NVIDIA. Tiny open 30B-A3B model with the lowest first-token latency in the index.
- Nemotron 3 Ultra — NVIDIA. NVIDIA’s largest open model: hybrid Mamba-Transformer MoE, 55B active.
- Nemotron 3 Embed 1B — NVIDIA. Small open embedding model for high-throughput enterprise retrieval.
- Cosmos 3 — NVIDIA. NVIDIA’s open omni-modal world foundation model for physical AI.
- Nemotron 3.5 Content Safety — NVIDIA. Compact 4B multimodal guardrail model for LLM and VLM traffic.
Tools that use it
- Isaac Lab — NVIDIA. GPU-accelerated robot learning framework on Isaac Sim.