RawMind

Connect your model

RawMind doesn't host any AI model itself. It talks to a model running on your own machine (or a server you control) through Ollama. Set that up once below, and every RawMind persona will use it.

Setting up Ollama

1. 1. Install Ollama

   Download it for your OS from ollama.com/download and install it like any regular app.

2. 2. Pull an uncensored model

   Open a terminal and run:

   ollama pull dolphin-mixtral

   Smaller machine? Try dolphin-mistralor a quantized variant instead — check the model's page on ollama.com/library for size/quality tradeoffs.

3. 3. Make sure Ollama is running

   It usually starts automatically after install and listens on http://localhost:11434. You can check with:

   curl http://localhost:11434/api/tags

4. 4. Point RawMind at it

   If this website and Ollama are on the same computer, use http://localhost:11434 above. If the website is deployed elsewhere (e.g. Vercel) and Ollama stays on your own machine, expose it with a tunnel first, then paste that URL instead:

   ngrok http 11434

   Then use the https://...ngrok-free.app URL it prints. Anyone with that URL can reach your Ollama server while the tunnel is open, so only share it with people you trust and close the tunnel when you're done.