Deploy a model to devices

Deploying puts a model to work on real devices. Where promoting a champion sets your project’s production model across the fleet, Deploy lets you push a specific model to a specific group of devices.

​

Deploy a model

1. Open Models and select Deploy on the model you want to ship.
2. Choose the group to deploy to — the model rolls out to every device in it.
3. Confirm. Each device picks up the model on its next check-in.

Detection labels come straight from the dataset’s classes — no extra setup.

​

Watch it run

Open Video to see live inference on your devices. Detections appear in real time, with Good / Defect events listed alongside the stream.

​

You’ve completed the loop

From here, collecting more data and retraining follows the same path — that’s how you adapt a model as the real world changes.