OpenUI is an open standard and framework for generative UI — letting AI models and agents respond with real interactive interfaces instead of walls of text. Created by Thesys, the project has earned more than 12,700 stars on GitHub under the MIT license and was in active development in 2026. More than 50,000 developers use it to give their AI apps dashboards, charts, forms and tables generated on the fly.
At the heart of OpenUI is OpenUI Lang, a compact, streaming-first language that LLMs write to describe interfaces built from your own components. Because each line defines one component, the UI renders progressively while the model is still generating — and in Thesys’s benchmarks it uses over 50% fewer tokens than JSON-based alternatives. First-party runtimes render the same response in React, Vue, Svelte, React Native or plain HTML, and you bring your own design system, so generated screens look like your product, not a generic chatbot widget. A CLI can even teach coding agents such as Codex, Cursor and Claude Code how to build with it.
OpenUI itself is free and open source under the MIT license. Thesys also operates OpenUI Cloud and the hosted Gateway/Observability services for production deployments, which are commercial offerings on top of the free framework.
OpenUI is for product teams and developers building AI copilots, agents and chat experiences who want model responses presented as rich, interactive UI — and who want an open standard rather than a proprietary widget format.