Cloudflare Workers AI

Cloudflare Workers AI runs open machine learning models on Cloudflare's serverless GPU network. It supports chat and text generation, embeddings, image and speech models, called from Workers or a REST API.

Website Docs Changelog Compare with others

Cloudflare Workers AI at a glance

From official pages, checked Oct 5, 2026. Tap the arrow for each source.

Free tier: 10,000 Neurons per day
Included on both the Workers Free and Paid plans; going beyond the daily allocation needs the Paid plan.
Starting price: $0.011 per 1,000 Neurons
Billed beyond the free daily allocation. Gemma 4 26B: $0.10/M input, $0.30/M output tokens.
Rate limit: 300 requests/min (text generation)
Default for text generation models unless the model requires the Workers Paid plan.
Trains on your data: No
Cloudflare does not use your Customer Content to train models on Workers AI without explicit consent.

What you can build with Cloudflare Workers AI

Cloudflare Workers AIRecipeConfig

Run an open chat model from a Cloudflare Worker with Workers AI

Call Google's open Gemma 4 model from a TypeScript Cloudflare Worker using the Workers AI binding. No API key, free daily allocation.

Free tier: 10,000 Neurons per dayFrom $0.011 per 1,000 Neurons

Chat

Cloudflare Workers AI at a glance

What you can build with Cloudflare Workers AI

Run an open chat model from a Cloudflare Worker with Workers AI

Recent updatesAll updates