Python on Vercel is now generally available

General availability means Vercel now supports building and running Python applications with the frameworks and tools developers already use. A deployed project is inspected for its dependency files, including projects managed with uv, and for frameworks such as FastAPI, Flask, Django and FastHTML; Vercel then configures the build, request routing and, where the framework supports it, static assets served from its CDN. The application runs on Fluid compute, Vercel's execution model that handles concurrent requests and scales with traffic (vercel.com).

Two design choices explain how this is meant to fit existing Python code. The runtime is CPython, the standard interpreter, on Linux, which Vercel chose to keep compatibility with the packages and tools Python developers already use. And because many Python backends sit behind a JavaScript frontend, Vercel Services puts both in one project: running vercel dev starts them together, and a preview deployment lets a backend change be tested against the frontend that calls it. Work that does not fit the request-response pattern, such as workers pulling jobs from a queue or database, is handed to Vercel Daemons, which keeps containers running and was released as a public beta alongside this announcement.

The other half of the launch is a set of Python SDKs for Vercel's own infrastructure, written to use type annotations, async and await, and context managers. They cover calling and streaming AI models and building agents, running code in isolated sandboxes, durable workflows with retries and saved state, queues for background messages, Blob file storage, and obtaining credentials for external APIs. Vercel demonstrates them in an open-source notebook app, vercel-labs/notebook-factory, which pairs a Vite frontend with a FastAPI backend and PostgreSQL. The post does not discuss pricing.