Open WebUI (github.com)

Open WebUI is a self-hosted AI interface for people who want to run a chat UI entirely offline rather than depend on a cloud chatbot. It connects to local models through Ollama or to local and cloud models through OpenAI-compatible APIs.

A user installs it with pip and runs it with open-webui serve, or runs it as a Docker container, then opens it in a browser at localhost. Different Docker image tags offer a standard image with local AI dependencies, a lightweight slim image that uses external services for embeddings and voice, CUDA acceleration, or a build bundled with Ollama, and a native desktop app is available for macOS, Windows, and Linux.

Its ecosystem includes Open Terminal, an action layer and enterprise orchestration that turns Open WebUI into a full agent harness with per-user workspaces; Knowledge Base Sync, which mirrors folders, repos, buckets, and wikis into Open WebUI Knowledge Bases with incremental sync; and Open WebUI Computer, which exposes a real machine's files, editor, terminal, and git inside the browser with optional AI agents.

Open WebUI is created by Timothy Jaeryang Baek and distributed under the Open WebUI License, which requires preserving Open WebUI branding, alongside prior contributions under their original licenses. It has more than 157.7k GitHub stars and 22.9k forks, and offers Enterprise and sponsorship options.

Use cases

• Running a self-hosted chat UI offline with local models
• Syncing an organization's documentation into a knowledge base