Run a model in your browser

Goal

Run a model entirely inside your browser with zero install, so inference happens on your machine without the Local Runtime.

What you need

• A modern browser with WebGPU support (planned).
• A model from the catalog that is prepared for in-browser inference (planned).

Steps

1. Open the Web App and switch the compute target to Browser (planned).
2. Pick a browser-compatible model from the model picker (planned).
3. The model downloads and is cached in browser storage (planned).
4. Chat runs fully in the browser; nothing is sent to BotConnector cloud (planned).

Expected result

A working chat that never leaves your browser, after a one-time model download into browser cache.

If it fails

Nothing to troubleshoot yet — the feature does not exist. When it ships, browser-specific guidance will appear here. Until then, local AI on your device works today: Run a model on your device. General problems: Troubleshooting.

Next step

Use what exists today: Use Cloud AI or Run AI on your device. Watch the status page for the browser capability to move from Planned.