OpenUI
OpenUI is an MIT licensed framework for rendering model generated forms, charts, tables and other interactive interfaces inside applications you control
Overview
OpenUI is an MIT licensed framework for letting AI applications return interactive interfaces rather than plain text. A model writes structured OpenUI Lang, and your application renders it as registered components such as charts, forms, tables and cards. The output appears progressively while the model is still responding.
You can use OpenUI with React, Vue, Svelte or Angular, and integrations are available for LangChain, LangGraph, assistant UI and CopilotKit. Developers define which components and tools the model can access, generate the corresponding system prompt, and connect an OpenAI compatible model endpoint. Local models can run through Ollama or LM Studio, so neither the framework nor your application requires a Thesys account.
This is a development framework, not a ready made server application. You will need JavaScript experience and an existing AI product or agent to use it. Smaller local models can also produce malformed interfaces when the component prompt exceeds their context window. OpenUI Gateway can validate and repair output, but it is an optional hosted service with free and paid plans. Hosted observability is also optional. Applications that expose queries, mutations or actions must validate permissions on the server because model generated parameters should never be trusted by themselves.
Use Cases
Specific ways to use OpenUI for your workflow.
Deployment Strategy
Recommended ways to host OpenUI in your own environment.