← Back

Cloudflare AI Worker

Infrastructure TypeScript Cloudflare
🌐

Problem

AI APIs are slow for users far from the provider's servers. A user in Tokyo hitting OpenAI's US servers sees 200ms+ of network latency before processing even starts.

Solution

An edge-deployed API gateway running in 300+ Cloudflare locations worldwide.

  • Low latency: Connect to nearest edge location
  • Multi-provider: Route to OpenAI, Claude, or Cloudflare AI
  • Rate limiting: Built-in abuse protection
  • Caching: Instant responses for repeated queries

Demo

Deploy the worker. Hit the /chat endpoint from different global locations. Compare latency vs hitting OpenAI directly. Show rate limiting kicking in after threshold.

Deploy to edge
Compare latency

Run it

git clone https://github.com/freshveejay/cloudflare-ai-worker
cd cloudflare-ai-worker
npm install
wrangler dev
# Deploy: wrangler publish