Documentation
Run AI in your own environment
Local deployment depends on the release and host platform. Choose the path that matches your setup.
Apple Silicon desktop
v0.0.24 includes a local AI engine without Docker or an external GPU. Prepare the model in the app, then test representative cameras for speed and memory use.
Windows, Linux, and your own GPU
The current desktop installers do not bundle a local GPU engine. Connect a model endpoint you provide. NVIDIA hosts can optionally run a vLLM container; that Docker path is separate from the bundled macOS engine.
Headless API deployment
For integration with your existing systems, request the API deployment kit and run it in your own environment. Check model and GPU requirements in the deployment guide before testing capacity.
Inference images stay in your environment when you use a local or controlled private model. Selecting a remote model sends sampled images to the configured endpoint.
Next
Remote GPU