LingBot-VLA 2.0

Robbyant

A vision-language-action release with an action representation that includes dexterous hands and other robot subsystems.

What you can explore

Explore model resources, custom dataset preparation and robot-specific post-training and inference configurations.

Before you use it

Expanded action representation is not proof of support for every robot. Verify the embodiment configuration and third-party model terms; training has substantial compute requirements.

Original contributors

Robbyant / LingBot-VLA contributors

License / access: Code: Apache 2.0 · check external data, models and hardware terms separately

Original code & documentation

Research paper

Editorial reference · checked 2026-09-22