Skip to main content

Open WebUI

What is Open WebUI?​

Open WebUI is a ChatGPT-style web interface for local models. Ollama gives you an API; Open WebUI gives you the chat window in front of it: pick a model, hold a conversation, upload a document to ask questions about, all in the browser with no cloud account and no per-token bill.

Start Open WebUI​

Open WebUI is a front end, so it needs a model backend running behind it. By default that's Ollama:

./laradock start open-webui ollama

Then open http://localhost:8322.

On first visit you'll be asked to create an account. It's stored locally in the container's own volume, nothing is sent anywhere.

Pull a model first​

A fresh Ollama has no models, so the model dropdown will be empty until you pull one:

docker compose exec ollama ollama pull llama3.2

Refresh Open WebUI and llama3.2 appears in the model picker. See Ollama for more on managing models.

Stop Open WebUI​

./laradock stop open-webui

To remove the container:

./laradock remove open-webui

Your accounts, chat history, and settings live in the open-webui Docker volume, so removing the container keeps them: starting it again picks up where you left off.

Configuration​

All settings live in open-webui/defaults.env and can be overridden by adding the same line to your own .env:

VariableDefaultWhat it does
OPEN_WEBUI_VERSIONmainImage tag from ghcr.io/open-webui/open-webui.
OPEN_WEBUI_PORT8322Host-side port Open WebUI is published on (container port 8080).
OPEN_WEBUI_LLM_ENGINEollamaWhich Laradock model service to talk to.
OPEN_WEBUI_LLM_PORT11434Port that service listens on.
OPEN_WEBUI_AUTHtruefalse skips the login screen entirely.

Point it at LocalAI instead​

OPEN_WEBUI_LLM_ENGINE is the service name Open WebUI talks to, so it retargets without editing any compose file. To use LocalAI instead of Ollama, set both the engine and its port in your .env:

OPEN_WEBUI_LLM_ENGINE=localai
OPEN_WEBUI_LLM_PORT=8080

Then start the pair:

./laradock start open-webui localai

Skip the login screen​

Signing in on every fresh volume gets old on a dev machine. Setting OPEN_WEBUI_AUTH=false in your .env drops the account system entirely and opens straight into the chat:

OPEN_WEBUI_AUTH=false

Only do this when nothing else can reach the port. With auth off, anyone who can open http://localhost:8322 gets full access, including any machine on your network if the port isn't firewalled.

Decide before the first boot, because the switch sticks in both directions: once an account exists Open WebUI refuses to turn auth off, and once you've run without it, turning it back on locks you out of the account it created for you. Either way the way out is a fresh volume:

docker volume rm laradock_open-webui

Change the Open WebUI version​

Set the version in your .env:

OPEN_WEBUI_VERSION=v0.5.20

open-webui runs a pulled image, not a local build, so pull the new tag first:

docker compose pull open-webui

Then recreate the container on it:

./laradock start open-webui

Common issues​

  • Model dropdown is empty. Ollama has no models until you pull one: docker compose exec ollama ollama pull llama3.2.
  • "Server connection failed". The backend isn't running. Start it (./laradock start ollama) and confirm OPEN_WEBUI_LLM_ENGINE matches a real Laradock service name.
  • Blank page or slow first load. The image is large and the first boot builds its local database. Watch progress with ./laradock logs open-webui.
  • Replies are very slow. The model runs on CPU unless a GPU is wired up, and larger models are slower. Try a smaller one (llama3.2:1b) or see Ollama for GPU notes.
  • Port 8322 already taken. Change OPEN_WEBUI_PORT in your .env.

Need the model engine itself? See Ollama or LocalAI. Routing to cloud providers too? LiteLLM puts one OpenAI-compatible endpoint in front of everything. New to Laradock? Start with Getting Started.