Nvidia unveils GR00T 2: an open AI model that teaches robots delicate hand skills from human video
Nvidia’s GR00T 2 teaches robots delicate skills from human video
Nvidia announced GR00T 2, an open world action model trained on robot data and 100,000 hours of first-person human video to help robots learn delicate hand skills. In Nvidia's tests, the model achieved 95% average success with five demonstrations per task, though those tasks were selected from ones the model could already learn well. The company released the training data for public use and noted that partners such as Foxconn, Siemens, and Nexuni are testing the model for AI server assembly, factory work, and laundry respectively, without confirming production deployment.
Source: Robotics — Robot learning news discovery · Read original article ↗
Article text · Original source · English
Nvidia unveiled GR00T 2, an open AI model that helps robots learn delicate hand skills, trained partly on 100,000 hours of human video. It reached 95% average success with five demonstrations per task in Nvidia’s tests. Foxconn, Siemens and Nexuni are testing it.
Nvidia has unveiled GR00T 2, an AI model that helps robots learn delicate hand skills partly by watching videos of people, the company said on Monday. It announced the model at the Conference on Robot Learning in Austin, Texas.
Nvidia calls GR00T 2 an open “world action model,” one that predicts how its actions will change what it sees before it moves.
It was trained on robot data and human video, including 100,000 hours of first-person footage that Nvidia is releasing for anyone to use. That lets robots learn skills from people and pass them between robots with different bodies, Nvidia says. Startups such as Maxinsights pay people to film their chores for the same purpose.
In Nvidia’s tests, GR00T 2 did 40 of 50 tasks at least once with no training for the task. Shown a task once, it succeeded 71% of the time on average, and 95% when shown five times. Nvidia’s own paper notes those tasks were picked from ones the model could already learn well.
Foxconn, which now earns most of its revenue from AI servers, is using GR00T 2 to help build them. Siemens is testing it for factory work and Nexuni is trying it for laundry. None has said it will use it in production yet.
Nvidia also released Isaac Lab 3.0, its open-source tool for training robots in simulation. It says the tool trains up to 2.5 times faster on one test. Partners include Walden Robotics, which launched with $300M in July.
“To me, agentic robotics is the most exciting technological development since the internet in 1995,” said Ken Goldberg, a robotics professor at UC Berkeley who worked with Nvidia on two of the projects.
Earlier GR00T models already underpin robots in development, including LG’s humanoid due in 2027. Siemens has also tested an Nvidia-powered humanoid on an eight-hour factory shift.
Published November 9, 2026 - 3:32 pm UTC
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Data origin
- Human egocentric dataOpen source S3
Reported numbers
training_hours
100,000 hours
View original evidence
Nvidia unveiled GR00T 2, an open AI model that helps robots learn delicate hand skills, trained partly on 100,000 hours of human video. It reached 95% average success with five demonstrations per task in Nvidia’s tests. Foxconn, Siemens and Nexuni are testing it. Nvidia has unveiled GR00T 2, an AI model that helps robots learn delicate hand skills partly by watching videos of people, the company said on Monday . It a
Open source S2demonstrations_per_task
5 episodes
View original evidence
Nvidia unveiled GR00T 2, an open AI model that helps robots learn delicate hand skills, trained partly on 100,000 hours of human video. It reached 95% average success with five demonstrations per task in Nvidia’s tests. Foxconn, Siemens and Nexuni are testing it. Nvidia has unveiled GR00T 2, an AI model that helps robots learn delicate hand skills partly by watching videos of people, the company said on Monday . It a
Open source S2tasks_completed_no_training
40
View original evidence
erent bodies, Nvidia says. Startups such as Maxinsights pay people to film their chores for the same purpose. In Nvidia’s tests, GR00T 2 did 40 of 50 tasks at least once with no training for the task. Shown a task once, it succeeded 71% of the time on average, and 95% when shown five times. Nvidia’s own paper notes those tasks were picked from ones the model could already learn well. Foxconn, which now earns most of
Open source S4factory_shift_hours
8 hours
View original evidence
agentic robotics is the most exciting technological development since the internet in 1995, ” said Ken Goldberg, a robotics professor at UC Berkeley who worked with Nvidia on two of the projects. Earlier GR00T models already underpin robots in development, including LG’s humanoid due in 2027 . Siemens has also tested an Nvidia-powered humanoid on an eight-hour factory shift . Published November 9, 2026 - 3:32 pm UTC
Open source S6simulation_speedup_factor
≤2.5
View original evidence
its revenue from AI servers, is using GR00T 2 to help build them. Siemens is testing it for factory work and Nexuni is trying it for laundry. None has said it will use it in production yet. Nvidia also released Isaac Lab 3.0, its open-source tool for training robots in simulation. It says the tool trains up to 2.5 times faster on one test. Partners include Walden Robotics, which launched with $300M in July. “ To me,
Open source S5
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
nnounced the model at the Conference on Robot Learning in Austin, Texas. Nvidia calls GR00T 2 an open “ world action model, ” one that predicts how its actions will change what it sees before it moves. It was trained on robot data and human video, including 100,000 hours of first-person footage that Nvidia is releasing for anyone to use. That lets robots learn skills from people and pass them between robots with diff
Open source S3
Nvidia unveiled GR00T 2, an open AI model that helps robots learn delicate hand skills, trained partly on 100,000 hours of human video. It reached 95% average success with five demonstrations per task in Nvidia’s tests. Foxconn, Siemens and Nexuni are testing it. Nvidia has unveiled GR00T 2, an AI model that helps robots learn delicate hand skills partly by watching videos of people, the company said on Monday . It a
Open source S2
erent bodies, Nvidia says. Startups such as Maxinsights pay people to film their chores for the same purpose. In Nvidia’s tests, GR00T 2 did 40 of 50 tasks at least once with no training for the task. Shown a task once, it succeeded 71% of the time on average, and 95% when shown five times. Nvidia’s own paper notes those tasks were picked from ones the model could already learn well. Foxconn, which now earns most of
Open source S4
agentic robotics is the most exciting technological development since the internet in 1995, ” said Ken Goldberg, a robotics professor at UC Berkeley who worked with Nvidia on two of the projects. Earlier GR00T models already underpin robots in development, including LG’s humanoid due in 2027 . Siemens has also tested an Nvidia-powered humanoid on an eight-hour factory shift . Published November 9, 2026 - 3:32 pm UTC
Open source S6
its revenue from AI servers, is using GR00T 2 to help build them. Siemens is testing it for factory work and Nexuni is trying it for laundry. None has said it will use it in production yet. Nvidia also released Isaac Lab 3.0, its open-source tool for training robots in simulation. It says the tool trains up to 2.5 times faster on one test. Partners include Walden Robotics, which launched with $300M in July. “ To me,
Open source S5
Implications for data suppliers
RoboSignal interpretation and collection questions, not statements of buyer demand.
- Check whether the intended buyer needs human observations, hand or object annotations, or aligned robot actions. Human video alone does not provide robot commands.
- Compare the reported units and scope before using these quantities in a budget. Recording hours, sensor-hours and trajectories are different measures.
- Validate demand and acceptance criteria with a buyer before scaling. Publication, popularity and a research result do not establish a purchase commitment.
Source：Robotics — Robot learning news discovery · thenextweb.com