Cloudflare Workers AI
Serverless GPU inference on Cloudflare's global edge network.
About
Workers AI is Cloudflare's serverless inference platform for running machine-learning models on its global network of GPUs. Developers call open-source models (Llama, FLUX, Whisper, and others) with a few lines of code, with inference running close to users across Cloudflare's edge. It launched in September 2023 and went generally available in April 2024, with a free tier of 10,000 neurons per day.
It pairs with AI Gateway (request routing and observability), Vectorize (vector database), and one-click deployment from Hugging Face. Honest caveats: the catalog is curated open-source models only — no proprietary frontier models — there is no model-training infrastructure, and serverless inference means shared resources with potential cold starts on spiky workloads.
What it can do
- Serverless inference on GPUs across Cloudflare's edge network
- Open-source text, image, audio, and embedding models via one API
- Free tier of 10,000 neurons per day with pay-as-you-go beyond
Setup
Enable Workers AI in the Cloudflare dashboard or via Wrangler; call models through the Workers AI binding or REST API.
Prerequisites
A Cloudflare account; a Workers project for the binding path.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 78; Trending score 88/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.