Cloudflare
Cloudflare runs one of the world's largest edge networks. Its Workers AI service runs open models on Cloudflare's own GPUs, billed per use.
Overview
Cloudflare is a US-based connectivity cloud company, best known for its content delivery network, DNS and security services. Its network spans data centres in more than 300 cities worldwide, which puts compute close to most internet users.
For LLMs, Cloudflare offers Workers AI: a serverless way to run open models such as Llama, Mistral, Qwen and gpt-oss on GPUs inside Cloudflare's own data centres, billed per use. It's built to work with Cloudflare Workers and Pages, and pairs with AI Gateway for request caching, rate limiting, analytics and routing between model providers.
Cloudflare suits teams that already build on its developer platform, or that want low-latency inference close to their users without managing GPUs. It's less suited to running custom or fine-tuned weights beyond the LoRA adapters it supports.
Plans from Cloudflare
| Hosting plan | Hosting type | Pricing range | Model families |
|---|---|---|---|
| Cloudflare Workers AI | Serverless API | From US$0.045 | Llama, Mistral, Qwen, gpt-oss |