Libre WebUI

Libre WebUI is a self-hosted AI workspace for individuals and teams who want chat, document research and AI agent tasks on infrastructure they control. You can use local models or connect a remote provider of your choice.

• Model choice: Chat through Ollama, llama.cpp or MLX LM on Apple Silicon, or connect OpenAI, Anthropic, Google Gemini and OpenAI-compatible services; compare model replies and fork conversations.
• Cited document answers: Search PDFs, Office files, Markdown, code and other documents by exact terms and meaning, with citations pointing to the relevant page, slide, sheet or section.
• Creative work: Generate and preview websites, code and multi-file projects, with image generation and editing through supported provider connections, including ComfyUI.
• Agent workspaces: Give tasks persistent files, a terminal and live previews, watch an agent use a sandboxed browser desktop, take control for sign-ins and turn demonstrations into reusable playbooks.
• Team sharing: Share resources with users or groups, use channels with model replies governed by the requesting member's access, and assemble reusable assistants with instructions, tools and knowledge.

The Apache 2.0 project runs on your machine, server or cluster, with desktop and Docker options. Local inference needs no cloud account. Local-provider prompts and responses remain where that provider runs; remote providers receive requests that can include documents and tool results. The app ships without application telemetry or analytics, with optional telemetry export.

Tools start disabled, and actions that change state require approval. Local Work sandboxes require Docker and have outbound internet access. The default Docker Compose setup gives the app root-equivalent host control through the Docker socket; a socket proxy or Kubernetes sandboxes offer stronger isolation.