NVIDIA/Isaac-GR00T

Health v2 · maintenance only

Adoption not part of the score

8116 stars · 1468 forks observed · 2026-09-23

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

NVIDIA Isaac GR00T N1.7 is an open vision-language-action (VLA) foundation model for generalized humanoid robot skills, taking language and image inputs to perform manipulation tasks across robot embodiments. The repository provides model weights, inference, fine-tuning, and evaluation tooling, with LeRobot integration.

Use cases

• run a vision-language-action model on a humanoid robot
• fine-tune a robot foundation model on custom demonstration data
• control robot arms for grasping and object manipulation from language commands
• evaluate cross-embodiment robot policies
• prototype generalist robot manipulation skills
• post-train a VLA model for a specific robot embodiment

When to choose

• you need a pre-trained VLA model for humanoid or bimanual manipulation
• you want to fine-tune a robot foundation model with your own demonstration data
• you are doing physical AI research on cross-embodiment policies
• you want LeRobot-compatible robot learning workflows

When to avoid

• you need a lightweight model for embedded microcontrollers
• you lack GPU hardware for inference or fine-tuning
• your task is unrelated to robot control or manipulation
• you need a non-neural classical robotics control stack

Facets

library · maturity stable

machine-learning deep-learning llm-inference sdk robotics artificial-intelligence machine-learning deep-learning python cross-platform vision-language-action foundation-model humanoid-robots robot-manipulation cross-embodiment fine-tuning lerobot nvidia gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="NVIDIA/Isaac-GR00T")