Real-robot experiment
Evaluated onRobot test results
NVIDIA reports G1 post-training and research evaluation of GR00T N1 against N1.5.
Scope: Unitree G1 research setup and tabletop object manipulation; vendor-reported results, not independent verification or factory preinstallation.
Conditions: G1 teleoperation data is used for post-training. The source reports N1.5 outperforming N1; it does not establish equal performance or commercial deployment.
Physical verification
We post-train GR00T N1 and N1.5 on 1K teleoperation episodes collected on the Unitree G1 robot. As in the GR-1 language following experiment, we initialize the scene with one target object and one distractor object, with the target object having equal probability of being closer to either the left or the right hand.
We observe that the post-trained GR00T N1.5 achieves much higher success rate than N1 for previously seen objects (toy fruits seen in the GR-1 pretraining corpus), and also shows generalization to various previously unseen objects.
Verified on: 2026-10-23