Rollback

Use this if you need to stop using Local inference, temporarily or permanently.

1. Disarm Local detection

In camera settings, disarm detection before making any other change. This stops the camera from calling the gateway.

2. Switch to Cloud inference

Switching to Cloud is an explicit action you take in camera settings. It is never done automatically on a Local failure. Confirm your Cloud setup is valid (a valid Hosted EDGE cloud configuration) before switching, then select Cloud as the inference mode and rearm detection.

3. Stop the gateway, if you are done with it

$HOME/.local/bin/anava-local-gateway service stop --state-dir "$HOME/.config/anava-local"

This stops the service without deleting anything. Your configuration, identity, certificate, and connection file remain in place, so you can restart it later without repeating setup.

4. Fully remove the gateway, if needed

$HOME/.local/bin/anava-local-gateway service uninstall --state-dir "$HOME/.config/anava-local"

This removes the launchd or systemd service definition only. It retains your gateway configuration, identity, certificate, and connection file under ~/.config/anava-local. Delete that directory yourself if you want to remove everything.

5. Release model memory, if needed

Stopping or uninstalling the gateway does not unload the model from Ollama.

ollama stop gemma3:4b

Recovery

To resume Local inference later, restart the service (or reinstall it), run a fresh sample detection to confirm the model is warm and correct, then rearm from camera settings. See Operate.

Related: anava.ai/edge/local