Open WebUI

Open WebUI is a self-hosted AI interface that connects to Ollama and OpenAI-compatible APIs. It is a front end, not a model by itself: you need a local inference server or a cloud provider to generate responses.

Setup options

Run the server with Docker or Python and open it in a browser. A separately versioned Desktop app is also available for Windows, macOS and Linux, with optional llama.cpp local models. Main application version: 0.13.0 (November 3, 2026). Desktop version checked: 0.1.4 (November 2, 2026). The platform buttons lead to the official project, where the Desktop downloads and server instructions are linked.

Useful features

• Keep multiple model connections behind a common chat interface.
• Organize chats and use workspace features.
• Use temporary side chats with the required permission; it is off by default.
• Connect optional Open Terminal integrations for terminal, browser and desktop workspaces, subject to the selected terminal's capabilities.

Privacy and costs

A local installation can keep inference local only when the model server and chosen features are local too. Downloads, cloud APIs, search and optional tools can contact external services. Check each connection before uploading sensitive documents. The software has no required subscription for normal self-hosted use; hardware, hosting and model providers can cost money.

License

Current releases use the Open WebUI License with a branding condition, plus earlier contributions under their original licenses. It is not simply MIT or BSD. Branding generally must remain intact; exceptions include deployments with no more than 50 end users in a rolling 30-day period, specific written permission or an enterprise license allowing changes. Read the license before redistribution or rebranding.

Screenshot: official Open WebUI animated demo, not a performance test. Read our Open WebUI discovery.

Sources: repository, 0.13.0 release, Desktop 0.1.4, license.