A live, clickable tour of Cloudflare Workers — from the traditional edge primitives to the 2026 platform & AI features. Every card below makes a real request.
Every request carries a cf object — country, city, datacenter, ASN.
Streaming HTML transformation — inject & rewrite content at the edge.
Global low-latency key-value store.
Serverless SQLite database, queried straight from the Worker.
S3-compatible storage, zero egress fees.
Strongly-consistent global counter — one object instance worldwide.
Programmatic edge caching — watch HIT vs MISS.
Scheduled execution — this Worker runs every 2 minutes.
Real-time fan-out over WebSockets. Open this page in two tabs.
Run Llama 3.1 on Cloudflare's GPU network.
Turn text into a 768-dim vector with bge-base.
Retrieval-Augmented Generation: embed → vector search → grounded answer.
Programmatically deploy isolated Workers into a dispatch namespace (Workers for Platforms), then invoke them on demand.
An isolated container — real shell, filesystem, processes — for AI agents to run code safely.