VLA models
SmolVLA
United States
A 450M open VLA trained only on crowd-sourced community data, small enough to train on a single consumer GPU and run on CPU or a MacBook; asynchronous inference yields ~30% faster response and ~2x task throughput. Built by Hugging Face's Paris-based LeRobot team.
OrganizationHugging Face (LeRobot team)
CountryUS
Release2025-06
Params (B)0.5
Open weightsyes
Architecturecompact VLA: SmolVLM-2 backbone + flow-matching action expert conditioned on multi-camera views, robot state and language; asynchronous inference
LicenseApache-2.0
Modalitiesvision, language, action
Embodimentsmanipulator
Training dataExclusively publicly available, crowd-sourced LeRobot community datasets from the Hugging Face hub (low-cost robot data)
Sources : arXiv (2025-06-02)Hugging Face (2025-06-03)Hugging Face (model card) (2026-07-09)
← Back to comparator · VLA models