Open WebUI

Open WebUI gives a household or company its own browser chat front end for LLMs, wired to a local Ollama server, to any OpenAI-compatible endpoint such as vLLM or OpenRouter, or several at once for side-by-side answers. It also bundles document retrieval with hybrid BM25 and vector search over nine vector stores such as Chroma, PGVector, and Qdrant, web search through SearXNG, Brave, Kagi, and many more, voice calls via local Whisper or hosted speech services, and image generation through ComfyUI or AUTOMATIC1111. Admins get role and group permissions, LDAP, OAuth, SCIM provisioning, usage dashboards, and a blind model comparison arena. Python plugins called filters, pipes, actions, and tools, plus MCP and OpenAPI tool servers, extend what models can do. Install it with pip on Python 3.11 or run the Docker image, whose tags can bundle Ollama or add CUDA. Code added since April 2025 uses the Open WebUI License: BSD-3 terms plus a rule against removing its branding in deployments above 50 users without permission or an enterprise deal; earlier code keeps its BSD-3 or MIT terms. Suits homelabs and firms wanting a private chat portal.