Gradio

Gradio turns a Python function into a web interface by mapping its arguments and return values onto UI components, so a model demo needs no HTML or JavaScript. The quickest route is gr.Interface, which builds a form from lists of inputs and outputs; gr.ChatInterface does the same for chatbots with streamed replies; and gr.Blocks exposes layouts, events and state for full applications, which is how large projects such as the AUTOMATIC1111 Stable Diffusion web UI were built. Dozens of components handle text, images, audio, video, files, dataframes and plots, and you can write custom components with a Svelte frontend. Every app also exposes an API that the gradio_client Python package or the @gradio/client JavaScript package can call, and passing mcp_server=True at launch lets AI agents use the same functions as MCP tools. Calling launch(share=True) opens a temporary public URL tunneled to your machine, and Hugging Face Spaces hosts apps free. Install with pip on Python 3.10 or newer; the 6.x series is current. Apache 2.0 licensed and maintained by Hugging Face, it suits ML engineers who need a usable front end fast.