Orange Book
Reference

Verifying plans, limits, and cost

Combine official limits, an observed request model, and budget alerts instead of treating a free tier as an architecture guarantee.

Edited and verified by Orange Book Editorial Team ·

REFERENCE · COST BOUNDARIES20 minutesVerified 2026-08-26

Quotas change

This page teaches verification, not permanent pricing. Open each product's current Pricing and Limits pages before every release and record the date.

Verified service facts

Current Workers Free boundary snapshot

Cloudflare can change plans and limits; reopen the primary source before implementation.

Product / planCurrent factScope and caveat
WorkersWorkers FreeDynamic requests100,000 requests / day

Resets daily at 00:00 UTC; static asset requests are excluded.

Verified · Source updated Primary source
WorkersWorkers FreeHTTP request CPU time10 ms / invocation

Time waiting on network, KV, or database I/O is not CPU time.

Verified · Source updated Primary source
WorkersWorkers FreeCompressed Worker size3 MB (gzip)

Use the gzip size reported by a Wrangler dry run.

Verified · Source updated Primary source

Build a minimum cost model

InputData neededOften missed
RequestsDaily peak, retries, static/dynamic ratioBots and health checks
CPUp50/p95 CPU rather than wall timeJSON parsing, SSR, cryptography
StorageCapacity, operations, retentionMigration copies and logs
DownstreamSubrequests, database queries, AI inferenceTimeout retries and batching

A budget is not one line of monthly requests times unit price. Expand a user action into Worker invocations, subrequests, storage operations, and inference, then apply peak and retry factors.

Release gates

Dry-run the bundle

pnpm wrangler deploy --dry-run --outdir .wrangler/bundle-check

Record gzip size. The current Workers Free ceiling is 3 MB; this project uses a lower internal threshold to retain upgrade room.

Define failure semantics

A security-critical Worker should not casually fail open. Decide whether bypassing the Worker or rejecting traffic is safer when request or CPU limits are reached.

Add observation and alerts

Before launch, assign an owner, threshold, and response for requests, errors, CPU, storage, and third-party spend.

Carry the cost model into each candidate

For object storage, bring public outbound into R2 vs S3, OSS, and B2. For databases, normalize rows, compute, and platform services before D1 vs Supabase and Neon. For deployment, continue with Workers and Pages vs other platforms.

Primary sources

Did this page help you complete your goal?

Beta feedback is generated in this browser and is never uploaded automatically.

On this page