Privacy

OwnLLM is built so that what you type stays on your computer.

• No accounts, no tracking, no analytics. There is no code in OwnLLM that reports usage, crashes or content anywhere.
• Your conversations are not saved by OwnLLM. The terminal commands keep nothing. The web page may keep a history in your own browser; clear it from the page's settings.
• The server stays on this computer. It listens only on 127.0.0.1. API and model requests need your private key (stored in ~/.config/ownllm/api.key, readable only by you); only the chat page itself and a simple health check answer without it.
• When OwnLLM uses the internet: only to download its own files - when you run the installer in download mode, or ownllm get <model> (the model files come from this project's Hugging Face page). Your prompts and usage are never sent. Each downloaded file is checked against a fingerprint before use. The one exception you control: if you run ownllm search --web or set a SearXNG address, the words you search for are sent to that service.
• Offline use: download the package and a model file once, and OwnLLM never needs the internet again.

What the AI itself might get wrong

A language model can be confident and mistaken. Do not rely on it alone for medical, legal, financial or safety decisions.