FLock launches a VLA robotics training task

matlabulous · x · 2026-07-20

FLock says it is expanding into robotics by launching a Vision-Language-Action (VLA) training task with FabricFND on FLock AI Arena.

The task trains a policy for a simulated 7-DOF Panda robot arm to:

The screenshot shows a live submission task for “VLA Models for Robotic Manipulation,” with a 9-day duration and an evaluation focused on task success, progress, and action quality. FLock frames this as a push toward open robotics through open model training.

Original post →

More from Embodied

Embodied channel →