NVIDIA unveils GR00T 2, a model that predicts robot video and actions together
NVIDIA says GR00T 2 predicts future video and robot actions together, reaching 40% zero-shot success on 50 unseen tasks and 95% after fine-tuning on five demonstrations per task.

Watch the short
Under a minute, read by an AI anchor. Made .
YouTube loads only when you press play.
NVIDIA introduced GR00T 2 on Monday. The robot model predicts future video and robot actions together. On 50 tasks that the target robot had not previously been shown, NVIDIA reports 40% average zero-shot success, meaning the robot attempted the tasks without any demonstrations for them. That figure rises to 95% after fine-tuning on five demonstrations per task.
The zero-shot number deserves a close read, because it measures performance before the model sees any examples of a new task. According to NVIDIA's figures, success was 71% with a single demonstration per task. The gains come in steps: about 40% with none, 71% with one, and 95% with five. NVIDIA describes the mechanism as the model predicting the next video frames and the robot's next actions together.
Code is planned for December, so outside researchers cannot yet run the model on their own hardware.
What changed
NVIDIA introduced GR00T 2, a robot model that predicts future video and robot actions together, with code planned for release in December.
Type: New capability · Applies to: GR00T 2
Why it matters
Robotics teams can see a path to deploying new robot tasks from a handful of demonstrations, though the code is not yet released.
What you can do
Add GR00T 2 to your evaluation list and plan testing once NVIDIA releases the code, which is planned for December.
Original source
We have no link to the original source for this story yet.
Where we saw it
Turing Post
AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works