Open WebUI

Open WebUI is a self-hosted web UI for large language models: multi-model chat with mid-conversation switching, RAG over uploaded documents, tools/functions, pipelines, web search, code interpreter, and voice input/TTS, with multi-user support and role-based access. It connects to Ollama, OpenAI-compatible APIs, or vLLM — the models themselves stay wherever you run them.

Deployment is a single Docker image (ghcr.io/open-webui/open-webui) in variants including a 196MB :slim image, :cuda for Nvidia GPUs, and :ollama with Ollama bundled. Important: it uses a custom, non-OSI "Open WebUI License" — not open source in the OSI sense. Removing "Open WebUI" branding in deployments with more than 50 end users in any rolling 30-day window is a material license breach without written permission or an enterprise license.

1. Install Docker. Generate a secret key: openssl rand -hex 32.
2. Run the container (verified against both the official docs and repo README):

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

1. Open http://localhost:3000 and create the first account — it becomes the admin. Do this before exposing the instance to a network.
2. Connect a model backend: point it at Ollama on the host (OLLAMA_BASE_URL=http://host.docker.internal:11434), or add OpenAI-compatible API keys in the admin panel.
3. For production, pin a version tag (e.g. :vX.Y.Z) instead of the rolling :main tag — the docs recommend this explicitly. Set OFFLINE_MODE=true on air-gapped instances to block Hugging Face traffic and the version check.