How to install Open WebUI with Docker and connect it to Ollama?

Updated: 28 September 2026 · Applies to: Open WebUI 0.11 with Docker on Ubuntu 24.04 / 26.04 LTS

Open WebUI is a self-hosted chat interface, similar to ChatGPT, that talks to your own Ollama server. You get accounts, chat history, document upload and a model picker in the browser, and your data stays on your server. This guide installs it with Docker on the same server that runs Ollama.

Before you start

Ollama is installed and answers: curl http://127.0.0.1:11434 prints Ollama is running (How to install Ollama on Ubuntu and run your first private LLM?).
Docker Engine is installed (Docker's Ubuntu instructions). Check with docker --version.
For GPU use nothing extra is needed here: Ollama itself uses the GPU (How to check that Ollama is using the GPU (and fix slow CPU-only answers)?).

Install Open WebUI

Create a secret key and copy the output. It keeps users logged in when the container is recreated:
```
openssl rand -hex 32
```
Start the container with host networking, so it can reach Ollama on 127.0.0.1 without opening Ollama to the network. Replace PASTE-YOUR-KEY with the key from step 1:
```
docker run -d --network=host \
  -v open-webui:/app/backend/data \
  -e OLLAMA_BASE_URL=http://127.0.0.1:11434 \
  -e WEBUI_SECRET_KEY=PASTE-YOUR-KEY \
  --name open-webui --restart always \
  ghcr.io/open-webui/open-webui:main
```
The image is large; the first start takes a few minutes. The open-webui volume holds all chats, users and settings, so never start the container without it.
Check that it runs:
```
docker ps
docker logs -f open-webui
```
Press Ctrl+C to leave the log view.
Open http://YOUR-SERVER-IP:8080 in your browser. With host networking the interface uses port 8080.
On the first screen, Get started with Open WebUI, click Create Admin Account. The first account becomes the administrator. Use a strong password. After that, new sign-ups are closed.
Pick a model in the model selector at the top. If it is empty, download one: type a model name (for example llama3.2) into the selector and choose the download prompt, or go to your avatar › Settings › Admin › Connections, click Manage next to the Ollama connection and download a model there.

Protect it

Port 8080 is now reachable from the internet if your server has no firewall. Either allow only your own IP address (sudo ufw allow from YOUR-IP to any port 8080 proto tcp, with sudo ufw allow OpenSSH first), or keep it closed and use an SSH tunnel: ssh -L 8080:127.0.0.1:8080 user@YOUR-SERVER-IP and open http://127.0.0.1:8080 on your computer.
For public use put a reverse proxy with HTTPS in front of it. Without HTTPS, passwords travel unencrypted.
To let other people register, go to your avatar › Settings › Admin › Authentication and turn on New Sign Ups. New accounts wait as Pending until you approve them.

Other ways to run it

Ollama on another server: use -e OLLAMA_BASE_URL=http://OTHER-SERVER:11434 and see How to make Ollama reachable from other machines (OLLAMA_HOST) safely? for making that server reachable in a safe way.
Everything in one container: the :ollama image variant contains Open WebUI and Ollama. With a GPU add --gpus=all (see How to use your NVIDIA GPU in Docker with the NVIDIA Container Toolkit?). Data goes into the volumes ollama and open-webui.
A fixed version: :main always points to the newest build. For production replace it with a version tag such as :v0.11.4 (see the releases page).

Frequently asked questions

The page does not open.
Check docker ps (status Up), your firewall for port 8080, and the container log.

The model list is empty or Open WebUI cannot reach Ollama.
See How to fix "Open WebUI cannot connect to Ollama"?.

How do I update it?
See How to update and back up Open WebUI without losing your chats?.

Official documentation: Open WebUI documentation: Quick Start (Open WebUI 0.11).

Want your own private AI without the setup work?

Private LLM installation: we install and configure Ollama, the GPU driver and a chat interface on your own server.
GPU dedicated servers: NVIDIA GPU servers for AI inference and training.
AI implementation services: private LLMs, RAG, n8n automation and API integrations built on your own servers.

Prefer a hand with the setup? Our engineers can do it for you: Hire an Expert, or use our on-demand server management.