The Pulse
Cloudflare Makes Python Workers Generally Available
Cloudflare has moved Python Workers into general availability, adding production support for FastAPI, Django, Flask, database drivers and AI libraries. The runtime connects Python applications directly to Workers AI, R2, D1, Hyperdrive, Dur
AI.info Team ·
Two years after introducing Python support in open beta, Cloudflare is making Python Workers generally available. The company says the release turns Python into a fully supported language on its Developer Platform, allowing applications to use Cloudflare services without JavaScript glue code.
“Today, Python Workers are now generally available (GA),” Gyeongjae Choi, Dominik Picheta and Hood Chatham write in Cloudflare’s announcement. The release supports familiar Python frameworks, database clients, AI libraries and package-building tools inside the Workers runtime.
From Pyodide experiment to supported runtime
Cloudflare first brought Python to Workers in April 2024 through Pyodide, a Python distribution compiled to WebAssembly. The initial release supported a limited selection of packages and required developers to work around the boundary between Python objects and the JavaScript-based Workers platform.
The general-availability release removes much of that friction. Python Workers can now use Cloudflare Developer Platform bindings directly through Python objects. A developer sending a dictionary to a Queue, for example, no longer needs to convert it into a JavaScript object with Pyodide helper code.
Cloudflare says the runtime and its Python software development kit now handle type conversion at the platform boundary. The change matters for ordinary developers as well as automated coding tools, because Python code can call services such as Queues and Durable Objects without requiring knowledge of the JavaScript environment underneath.
FastAPI, Django and Flask run through native connectors
Python Workers now include connectors for the Web Server Gateway Interface and its asynchronous counterpart, the Asynchronous Server Gateway Interface. Those interfaces allow existing Python web applications to run on Workers without replacing their framework-specific application code.
FastAPI applications can use the workers.asgi package, while synchronous applications such as Django projects can use workers.wsgi. Flask and other frameworks built around WSGI or ASGI can use the same approach.
Cloudflare’s runtime takes responsibility for handling incoming requests and scaling execution. The connector translates the request into the structures expected by the Python framework, then returns the framework’s response through Workers. Developers do not need to run Uvicorn, Gunicorn or another separate application server inside the deployment.
Hyperdrive brings PostgreSQL and MySQL drivers
Database access was one of the harder technical barriers for Python Workers. Many Python database libraries depend on the standard socket module, while WebAssembly environments traditionally lack the operating-system networking calls those libraries expect.
Cloudflare says it addressed the problem by implementing socket operations through the Workers Connect API. Python drivers can use standard socket calls, while the runtime translates those operations into the networking functions available inside Workers.
The change enables Hyperdrive integrations for relational databases including PostgreSQL and MySQL. Developers can use familiar libraries such as aiomysql and asyncpg, subject to the packages supported by Cloudflare’s Python Workers environment.
Package support moves beyond a Cloudflare-only toolchain
Python packages that include native C, C++ or Rust extensions must be compiled for WebAssembly before they can run inside Python Workers. Cloudflare previously built and hosted many of those packages itself, which limited the available selection and left maintainers without a common distribution target.
The company says it helped develop PEP 783, a standard for running Python packages on the PyEmscripten platform. The proposal was accepted after more than a year of discussion, giving package maintainers a shared target for Python builds intended for browser and other WebAssembly environments.
Cloudflare also says it stabilized the Pyodide build toolchain and added PyEmscripten support to cibuildwheel, a widely used tool for building Python packages. Adoption is still in progress, so general availability does not mean every package on the Python Package Index will work in Workers.
Python becomes an option for AI agents at the edge
Networking support also expands what developers can build with Python Workers. Cloudflare says libraries including openai, langchain and mcp can now run natively, after changes that allow Python HTTP clients to route requests through the JavaScript fetch API in WebAssembly environments.
Applications can combine those libraries with Workers AI for inference on Cloudflare’s GPUs or use AI Gateway to route requests to external model providers. Cloudflare’s examples include a LangChain application that calls a Workers AI model and returns the result from a Python Worker.
The same platform can connect that application to R2 storage, D1 databases, Queues, Workflows, Durable Objects and Vectorize. Cloudflare also presents Python examples for an image-to-image generation pipeline, a Bluesky Jetstream stream processor, an MCP server and a retrieval-augmented generation system.
General availability raises the bar for package compatibility
Cloudflare is updating its developer documentation so that Python examples appear alongside JavaScript and TypeScript examples across its products. The company says it plans to improve Python Workers’ performance and memory efficiency while adding support for more packages.
The release gives Python developers a production path into Workers, but the platform still depends on WebAssembly-compatible packages and Cloudflare-specific runtime behavior. Developers moving an existing application will need to check framework compatibility, database-driver support and any native extensions before deployment.
For projects that fit those constraints, Cloudflare is no longer presenting Python Workers as an experiment. The service now treats Python applications, AI pipelines and web frameworks as first-class workloads inside the Workers runtime.