Set up Open WebUI on a Mac for a private ChatGPT
Short answer
Install Ollama and a model, then run Open WebUI with "uvx --python 3.11 open-webui@latest serve" (or pip with Python 3.11 or 3.12) and open http://localhost:8080 in your browser. It finds Ollama automatically and gives you chat history, file uploads and model switching, with nothing leaving your Mac.
Ollama runs models in the terminal. Open WebUI adds a familiar chat interface in your browser: saved conversations, a model picker, document uploads and more. Everything stays on your Mac.
What you need
- Ollama installed, with at least one model downloaded (for example
ollama pull qwen3.5:9b). - Python 3.11 or 3.12. Open WebUI does not support Python 3.13 yet, according to its quick start guide.
Option 1: Install with uv (recommended)
uv is a fast Python tool manager that fetches the right Python version for you.
brew install uv
DATA_DIR=~/.open-webui uvx --python 3.11 open-webui@latest serve
DATA_DIR keeps your chats in a folder in your home directory, so they survive updates.
Option 2: Install with pip
If you already have Python 3.11 or 3.12:
python3.11 -m venv ~/open-webui-env
source ~/open-webui-env/bin/activate
pip install open-webui
open-webui serve
Open it
Go to http://localhost:8080 in your browser. The first account you create becomes the administrator. Open WebUI finds Ollama at http://localhost:11434 automatically, and your downloaded models appear in the model picker at the top.
Useful features
- Chat with documents. Drag a PDF or text file into the chat and ask questions about it.
- Compare models. Pick two models and send the same prompt to both.
- Download models from the interface. Type a model name such as
gemma4:26bin the model picker to pull it through Ollama. - Reuse prompts. Save prompts you use often and call them with a slash command.
Starting it again later
Run the same command you used to install (Option 1), or activate the environment and run open-webui serve (Option 2). Keep the Terminal window open while you use it.
Troubleshooting
- No models in the picker: make sure the Ollama app is running, then check Settings › Admin › Connections points at
http://localhost:11434. - Replies are slow: the model may be too large for your memory. See how much memory you need.
Frequently asked questions
Is Open WebUI free?
Yes. Open WebUI is free and open source. It runs on your own Mac and does not need an account with any cloud service.
Do I need Docker to run Open WebUI on a Mac?
No. Installing with uv or pip is simpler on a Mac and avoids running Docker in the background. Docker is an option if you already use it.
Can other devices on my network use it?
Yes. Other devices on the same network can open your Mac's local address on port 8080. Set a strong admin password first, since anyone who can reach that address can try to sign in.