Open WebUI
It was open source. Then the license changed.
What it is
A feature-rich self-hosted web interface for LLMs — chat, RAG, multi-user, pipelines, tool calling — working with Ollama and any OpenAI-compatible API, designed to run fully offline.
Why it's interesting
The highest-profile relicensing in local AI. It moved from permissive terms to a custom license that forbids removing the 'Open WebUI' branding unless you have 50 or fewer users in a rolling 30-day window (or written permission). That branding-protection clause is a use restriction OSI licenses don't allow.
Use cases
- A self-hosted ChatGPT-style front end
- Team AI portals over local or API models
Who it's for
Self-hosters, homelabs, small internal deployments
Setup
Easy. A single Docker container or pip install; an LLM backend
Limitations & cautions
You may not alter or remove the branding above the user threshold — a material-breach clause that takes it outside the open-source definition.
Editorial takeaway
A cautionary tale about depending on a license that can change under you. Still great software — for a fully open equivalent, look at Jan.