Price Cloudflare Workers on requests plus CPU milliseconds and check whether the free daily allowance still covers you.
The paid plan is a small flat fee that includes a block of requests and a block of CPU milliseconds; only usage beyond those blocks is metered. Because CPU time is billed separately from invocations, a Worker that does heavy JSON or crypto work can cost more in milliseconds than in requests even at modest traffic. Workers execute at whichever of Cloudflare's 300+ locations is nearest the user, so the same code replaces both a regional function and a CDN rule — but the CPU meter punishes slow code globally. Confirm the current plan fee, allowances and unit rates with Cloudflare before budgeting.
Cloudflare Workers Cost
cost = plan fee + max(0, requests − included requests) × request price + max(0, requests × CPU ms − included CPU ms) × CPU price.
cost = plan fee + max(0, requests − included requests) × request price + max(0, requests × CPU ms − included CPU ms) × CPU price. The paid plan is a small flat fee that includes a block of requests and a block of CPU milliseconds; only usage beyond those blocks is metered. Because CPU time is billed separately from invocations, a Worker that does heavy JSON or crypto work can cost more in milliseconds than in requests even at modest traffic.
Workers execute at whichever of Cloudflare's 300+ locations is nearest the user, so the same code replaces both a regional function and a CDN rule — but the CPU meter punishes slow code globally. Confirm the current plan fee, allowances and unit rates with Cloudflare before budgeting.
This calculator takes 8 inputs: Requests per month, CPU time per request, Workers plan fee, Requests included in the plan, CPU-milliseconds included, Price per million requests above the allowance, Price per million CPU-milliseconds above the allowance, Free-plan daily request allowance. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.
No. Only CPU time is billed, so a Worker that spends 200 ms waiting on an origin fetch but 3 ms computing is billed for roughly 3 ms. That is why the CPU-per-request input here is much smaller than observed response time.