Skip to content

Open WebUI

Chatting in a terminal is fine, but most people prefer a proper chat page. Open WebUI is an open-source web interface for your local models — it looks and feels like the big cloud chat sites, but it talks to your Ollama and your words stay on your machine.

What it gives you over the terminal: a familiar chat layout, conversation history, switching between models mid-chat, and a place to set system prompts without JSON. What it doesn't give you: the cloud models themselves. It's a nicer front door to the same local models.

Install concept

Honest label first: this was not installed on the writing machine, so treat the steps below as the documented concept, not a verified walkthrough. Check the project's current docs before running anything.

The documented install runs Open WebUI as a Docker container (Docker is a tool that runs packaged programs — installing Docker itself is a separate step, covered in Docker for Mere Mortals):

docker run -d -p 3000:8080 \
  -v open-webui:/app/backend/data \
  --name open-webui \
  ghcr.io/open-webui/open-webui:main

Then you open http://localhost:3000 in your browser, create a local account, and point it at your Ollama (it usually finds http://localhost:11434 on its own). From there: pick a model from the dropdown and chat.

What I can and can't promise here

Stated plainly
The concept Real: a web UI talking to your local Ollama API
The install command From the project's docs, not run during writing
Whether it works on your machine Depends on your setup — Docker, ports, and versions all matter

If the install fights you, the Ollama command line from Your First Model still works fine — Open WebUI is convenience, not a requirement. Other front-ends exist too; the idea is the same regardless of which one you pick: a local page, talking to a local API, running local models.

Worth knowing

  • Open WebUI keeps its data (chats, settings) in that -v volume — your conversations live on your disk, like everything else local.
  • It's another thing to update. New versions come when you pull them, same maintenance deal as the models themselves.
  • On a small machine (say, a Raspberry Pi), a web UI plus a model is a lot to ask of modest RAM — the terminal chat is lighter.

Check: after install, http://localhost:3000 shows a login page, and after signing in, the model dropdown lists the models from your ollama list. If the dropdown is empty, Ollama isn't reachable — see Troubleshooting.

Next: Choosing Models