Nvidia puts an open vision-language-action model on Hugging Face

theteknosaur · x · 2026-07-28

Nvidia has put an open vision-language-action model on Hugging Face for anyone to download and build on.

The poster frames it as a deliberate strategy: open-source the pieces that push more GPU demand. The post is short, but the signal is that Nvidia is pushing an open model release in the vision-language-action space, encouraging downstream builders to use and extend it.

Original post →

More from Infra

Infra channel →