Open WebUI is an self-hosted and extensible AI platform, designed to run fully offline. It provides a rich and user-friendly chat interface to interact with language models, whether served by Ollama or any OpenAI-compatible API. Full Markdown and LaTeX support, simultaneous multi-model conversations, image generation and editing, voice/video calls with multiple speech synthesis and recognition engines are part of the experience.

The tool integrates an embedded inference engine and a local RAG (Retrieval Augmented Generation) function compatible with nine vector databases, as well as web browsing and document integration via the “#” command. For developers, there are native Python function calls, a plugin framework (Pipelines), and image generation by multiple engines.

Also designed for enterprise, Open WebUI offers role-based access control, LDAP/Active Directory and SSO authentication, SCIM 2.0 provisioning, support for SQLite or PostgreSQL, OpenTelemetry observability, and Docker/Kubernetes deployment with horizontal scaling.