Run VLA models on real robots.
Pull a model, connect your hardware, and run it — without fighting mismatched environments, outdated SDKs, or a complicated setup.
curl -fsSL https://argos.sh/install | bashThe problem
Running an open-source VLA model on real hardware shouldn't mean fighting mismatched environments, hunting down outdated SDKs, and rebuilding a CLI setup every time something breaks.
You didn't start building a robot to debug dependency conflicts. You started to make it move.
What Argos is
A developer tool for running and connecting vision-language-action models on real robots.
Argos is built for robotics developers working with actual hardware, not simulation. Pull an open-source model, connect your arm and cameras, and run — one tool, one workflow, from your terminal.
What it does
Three commands between you and a working robot.
Pull
Download an open-source VLA model with a single command. Argos fetches and caches the weights for you — no manual downloads, no guessing where files should go.
Connect
Link your robot arm and cameras. Argos verifies each connection is actually working before you rely on it, not just that a port opened.
Run
Give it a plain-language instruction and watch your robot carry it out. Argos handles everything in between.
Argos supports SmolVLA, with more open-source VLA models on the way.
View the model catalogGet your robot running in minutes.
One install command. No environment wrangling.
curl -fsSL https://argos.sh/install | bash