Skip to main content

AI models that run on this device.

Two small chat models run in a computer’s browser here; every other model is a file you download.

What stays on your computer?

LEARN / LOCAL AI Sources checked

Local chat stays on your computer when you use a downloaded model without online tools.

Local and online actions

These are different actions even when they happen in the same app.

Local and online actions
ActionWhere it happens
Chat with a downloaded model, without online toolsYour computer runs the model.
Browse, search or download modelsThe app contacts the catalog or file host.
Use a cloud model, web search or a connected toolThat service receives the request.
Hugging Bay app fingerprint lookupThe app sends the file's SHA-256 when you ask.
Check for app or runtime updatesThe app contacts its update service.

Check your setup

  1. Download the app, runtime and model while online.
  2. Load the model and send a harmless test message.
  3. Disconnect the internet and send another message.
  4. Reconnect when you need to search, download or update.

An offline reply demonstrates that this chat can run locally.

Keep Ollama in local mode

In Ollama's ~/.ollama/server.json, set disable_ollama_cloud to true, preserving any other settings, then restart Ollama. Its FAQ also documents OLLAMA_NO_CLOUD=1. Avoid cloud models, web search and connected tools for a local-only chat.

Keep access limited

Leave local servers on their default loopback address. Keep exported chats and backups in a location you control. Use your app's delete controls when you want to remove a saved chat.

For site data, see Hugging Bay's privacy page.

Sources and checks

Checked 2026-10-17. Install requirements and availability can change; the linked source has current details.

Read as Markdown