In this post we will install openWebUI on Ubuntu 22.04 and configure it to act as the front end for Ollama (which is already running locally).

Note that we are not going to be using containers, rather we will install and run OpenWebUI as a service, much as we did with Ollama in a previous post.

Installation via Pip

Create virtual env.

 $ sudo python3.12 -m venv ~/Downloads/workspace/open-webui/

Install open-webui.

$ sudo ~/Downloads/workspace/open-webui/bin/pip3 install open-webui

Create systemd service file.

$ sudo vi /usr/lib/systemd/system/open-webui.service

See contents below. Note how ExecStart points to the open-webui binary in our virtual env.

Refresh systemd via the command below.

$ sudo systemctl daemon-reload

Now configure our new service to start at boot and enable it now.

$ sudo systemctl enable --now open-webui.service
Created symlink /etc/systemd/system/multi-user.target.wants/open-webui.service → /usr/lib/systemd/system/open-webui.service.

If you need are experiencing a failure or need to test, just run the ExecStart command manually and watch for errors.

$sudo /home/cpaquin/Downloads/workspace/open-webui/bin/open-webui serve

Open-WebUI should be listening on port 8080 on all interfaces. We can see this when we start it manually as we did above.

INFO:     Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)

Now restart via systemd to ensure everything is up and running nice and clean.

$ sudo systemctl status open-webui.service

---

Logging into the WebUI

Open a browser window and navigate to the IP address of your system. Or if running locally you can use the loopback address (127.0.0.1). Note you may need to allow port 8080 via ufw if attempting to access from a remote host.

Start by selecting “Get Started”

Create a local admin account

OpenWebUI should automatically detect your installation of ollama and allow you to load any downloaded model.

Testing with A Newly Installed Model

Lets pull a new model and make sure that OpenWebUI makes the new model available.

$ sudo ollama pull smollm

After refreshing our browser window, we can now see that the smolln model is available.