Setting up Open WebUI: a pleasant chat interface for your local models

When experimenting with local language models, the standard command-line interface of tools like Ollama is not always intuitive for daily use. Open WebUI (formerly Ollama WebUI) offers an excellent solution for this. It is a feature-rich, self-hosted web interface that strongly resembles commercial chat environments, but runs entirely locally while preserving your privacy.

What is Open WebUI?

Open WebUI acts as a user-friendly shell around your local or remote LLM backend. The application supports Ollama and OpenAI-compatible APIs out of the box. Key features include:

• RTE & Markdown: Full support for Markdown, code highlighting, and mathematical formulas (LaTeX).
• RAG (Retrieval-Augmented Generation): The ability to upload and query documents (PDFs, txt files) using local embeddings.
• Multi-user management: Support for multiple user accounts with role-based access control.
• Web Search Integration: Optional integration with search engines like SearXNG to retrieve up-to-date information.

Step 1: Preparing Ollama

Before starting Open WebUI, Ollama must be operational and have at least one model loaded. Install Ollama through the official channels and download a model of your choice via the terminal:

ollama run llama3

By default, Ollama runs on http://localhost:11434. Make sure this port is accessible to the container in which Open WebUI is running.

Step 2: Installing Open WebUI via Docker

The most reliable and simple way to deploy Open WebUI is via Docker. Run the following command to start the container:

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

This makes the web interface locally available on port 3000. The --add-host flag allows the container to easily communicate with Ollama on the host machine.

Step 3: First use and basic settings

After starting, navigate to http://localhost:3000 in your browser. On your first visit, you will be asked to create an administrator account. This account will have full permissions over the environment.

After logging in, Open WebUI automatically detects the models available via Ollama. In the settings menu, you can then:

• Fine-tune system prompts and parameters (such as temperature and context window) per model.
• Manage authentication settings for additional users within your network.
• Configure external API keys if you also want to connect cloud services.