The Stack

Cloudflare ships Python Workers GA with native AI bindings

Editorial

Editorial

— 3 min read
Share
Cloudflare ships Python Workers GA with native AI bindings

Cloudflare has spent about two years getting Python onto its edge network. As of Monday it is telling developers to build on it for real.

Cloudflare has moved Python Workers to general availability, making Python a first-class language on its developer platform — and, for anyone building AI, giving Python code native bindings to Workers AI, R2, D1, Hyperdrive, Durable Objects, Queues and Workflows instead of forcing conversions into TypeScript at the RPC boundary. That last detail is the one that matters. Until now, handing a Python dictionary to a Cloudflare Queue required hand-written glue code, and glue code is exactly the friction that kept agent and retrieval pipelines off the edge. FastAPI, Django and Flask now run inside Workers through a built-in connector, which means a Python service no longer needs a server process in front of it — the network handles the concurrency and scaling that Uvicorn or Gunicorn would otherwise do.

Under the hood it is all WebAssembly. Python Workers are built on Pyodide, so packages with native C, C++ or Rust extensions have to be cross-compiled before they can run in the sandbox, and Cloudflare previously hosted its own bespoke builds. The GA release switches that to a shared standard called PyEmscripten and works with package maintainers to publish WebAssembly-compatible wheels — the difference between "we support the packages we compiled" and "the ecosystem can build for us." The company has also extended its developer docs so that nearly every TypeScript example now has a Python twin.

The AI framing is not incidental. Cloudflare's launch material leads with MCP servers, retrieval-augmented generation against Vectorize, and asynchronous AI orchestration as the flagship Python workloads — our explainer on what RAG actually is covers the pattern it is courting. Database drivers reach Postgres and MySQL through a socket bridge implemented at the system-call level, so the drivers developers already use work unchanged. For a company that has spent the year positioning itself as the place agents run — it already builds edge tooling around frontier models, as in Cloudflare turns OpenAI's cyber model into an edge patch shop — getting production Python onto the same network closes an obvious gap in the pitch.

The honest caveat is cold starts and package coverage. Pyodide's history is a trail of packages that either would not build for WebAssembly or ran well behind their native versions, and Cloudflare lists better performance and memory efficiency as future work rather than shipped work. General availability is a promise about the platform, not about your dependency tree: if the library you depend on has no WebAssembly build, today changes nothing for you.

What to watch: whether the Python AI stack lands at the edge as fast as the TypeScript one did, and whether a steady stream of PyEmscripten wheels follows the standard rather than the announcement.

Would you move a Python AI service to the edge for the latency, or does the WebAssembly package ceiling keep you on containers? Tell us in the comments.

Read more

Mistral's Le Chonk puts Europe's sovereignty bet on a download date

Mistral's Le Chonk puts Europe's sovereignty bet on a download date

Mistral's biggest model ever is real, benchmarked and for sale today — but the thing that makes it matter to Europe's sovereignty argument, the weights, is still three weeks out. The preview settles who built it; the release will settle whether it counts. What Mistral actually shipped Mistral opened a public preview of Mistral Large 4 — unofficially ML4, officially le Chonk — a 1 trillion-parameter mixture-of-experts model with 49 billion active parameters and native multimodal input. The p

Mistral unveils Le Chonk: a 1T-parameter open-weights model

Mistral unveils Le Chonk: a 1T-parameter open-weights model

The biggest open-weight release outside China lands in public preview today, and the country that spent the week promising its own frontier model just put a price on the ambition. Mistral has opened a public preview of Mistral Large 4 — codenamed "le Chonk" — a 1 trillion-parameter mixture-of-experts model with 49 billion active parameters, natively multimodal, which the company calls its largest and most capable model to date. The preview API is live today on Mistral Studio at $1.36 per milli

Google signs 3.6 GW power deal, a quarter of it new nuclear

Google signs 3.6 GW power deal, a quarter of it new nuclear

Grid capacity, not chips, is becoming the binding constraint on the AI buildout — and on the same day, the labs told an Australian inquiry they can live with mandatory incident reporting. Google has contracted 3.6 GW of power from Constellation Energy across the PJM grid — the largest electricity deal in the region's history, and the biggest single power commitment any AI company has made. The agreement covers 3,590 megawatts over 13 states, with 890 megawatts of new nuclear capacity coming fr

Korea probes AI agents in bank hacks as president cites 'signs'

Korea probes AI agents in bank hacks as president cites 'signs'

South Korea opened a formal investigation into whether AI agents drove a wave of bank breaches — and it isn't the only AI story moving money today. President Lee Jae Myung said "signs" point at AI models, and the probe is now at the highest level a national banking sector has seen. Speaking at a cabinet meeting, Lee said that "in some hacking incidents, signs have emerged of AI being used, causing considerable public concern and anxiety," and police have since opened a full-scale investigation