How to Set Up Open WebUI for Local AI

Ensure Ollama Is Running

Before installing Open WebUI, verify that Ollama is installed and its service is running. Open a browser and go to http://localhost:11434. You should see a message like "Ollama is running." If not, install Ollama first using our installation guide, then return here.

Also confirm you have at least one model downloaded. Run ollama list in a terminal to see your installed models. If the list is empty, download a model with ollama pull qwen3:8b before proceeding. Open WebUI will work without models installed, but you will not be able to chat until one is available.

Choose Your Installation Method

Open WebUI supports three installation methods. Docker is recommended for most users because it handles all dependencies automatically and keeps Open WebUI isolated from your system. The desktop app is a good choice if you do not want to use Docker. Pip install gives you the most control but requires managing Python environments.

Docker: Best for most users. Requires Docker Desktop (macOS, Windows) or Docker Engine (Linux) installed on your system. If you do not have Docker, download Docker Desktop from docker.com and install it.

Desktop app: Best for users who prefer a native application. No Docker required. Available for macOS, Windows, and Linux.

Pip install: Best for Python developers who want direct control. Requires Python 3.11 specifically (not newer versions due to dependency constraints).

Install Open WebUI

Docker method: Open a terminal and run the following command:

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 command pulls the Open WebUI image, creates a persistent volume for your data (conversation history, settings, user accounts), maps port 3000 on your machine to the container, and configures host networking so the container can reach Ollama running on your host machine. The container restarts automatically when your computer boots.

Important for Mac users: Docker on macOS does not have GPU passthrough to the Metal framework. Run Ollama natively (outside Docker) and only run Open WebUI in Docker. The --add-host=host.docker.internal:host-gateway flag in the command above allows the Docker container to connect to your native Ollama instance.

Desktop app method: Download the Open WebUI desktop app from the project's GitHub releases page or website. Install it like any other application. On first launch, the app can either connect to an existing Ollama instance or manage its own built-in model engine.

Pip method: Ensure you have Python 3.11 installed, then run:

pip install open-webui

After installation, start the server with open-webui serve. The web interface will be available at http://localhost:8080 by default.

Create Your Account and Start Chatting

Open your browser and navigate to http://localhost:3000 (Docker) or http://localhost:8080 (pip install). You will see a signup page. The first account you create becomes the admin account with full control over settings, user management, and model configuration.

After creating your account and logging in, you see the chat interface. Select a model from the dropdown menu at the top of the page. The dropdown shows all models available through your connected Ollama instance. Pick a model, type a message, and press Enter to start your first conversation.

Conversation history saves automatically and appears in the sidebar. You can create new conversations, rename existing ones, search through past conversations, and delete conversations you no longer need. The interface renders markdown, highlights code blocks with syntax coloring, and supports copying responses with a single click.

Configure Settings and Preferences

Click the gear icon or your profile picture to access settings. The admin panel provides several useful configuration options:

Default model: Set which model loads by default when you create a new conversation. Choose your most-used model so you do not have to select it every time.

System prompt: Set a global system prompt that applies to all conversations, or create per-conversation system prompts for specific tasks. System prompts shape the model's behavior, tone, and focus.

Cloud API connections: Under the "Connections" settings, you can add OpenAI-compatible API endpoints. Enter your API key for OpenAI, Anthropic (via a proxy), or any compatible service, and those models appear alongside your local models in the model dropdown. This enables a hybrid local and cloud workflow from a single interface.

Multi-user setup: As admin, you can enable user registration so family members or team members can create their own accounts. Each user gets separate conversation histories, model preferences, and settings. You can control whether new users can register freely or require admin approval.

Document uploads: Open WebUI supports uploading PDF and text files directly into conversations. The system extracts text content and makes it available to the model as context. You can also create persistent knowledge bases using the RAG (retrieval-augmented generation) feature in the admin panel.