Back to Search
R

RT-2-X

Model

RT-2-X is a member of the RT-X model family, trained on the collaborative Open X-Embodiment Dataset. It extends the original RT-2 VLA model by fine-tuning on cross-embodiment data from 34 institutions, enabling generalization across diverse robot platforms.

Details

Updated:6/21/2026
open sourcefalse
release date2023-10-17
paper urlhttps://arxiv.org/abs/2310.08864
model familyRT-X

Tags

rt-x-familycross-embodimentgoogle-deepmindVLAgeneralist-policymulti-embodimentvision-language-action