Nvidia introduces GR00T 2 at CoRL

Nvidia said on Nov. 9, 2026, at the Conference on Robot Learning in Austin, that GR00T 2 is an open world-action model built on Cosmos, trained on robot and human data including 100,000 hours of Nvidia's open human egocentric video, and that the same post introduces Isaac Lab 3.0. The company said GR00T 2 completed 40 of 50 evaluated tasks at least once without task-specific demonstrations, with average success of 71 percent after one demonstration per task and 95 percent after five. Foxconn is applying it to server-factory automation, Siemens is evaluating it, and Nexuni is testing it for laundry handling.