Serverless Architecture Cost Estimator

AWS Lambda charges on two dimensions: number of requests ($0.20 per 1M requests) and duration (GB-seconds of memory × execution time). Duration is billed in 1ms increments at $0.0000166667 per GB-second. The free tier includes 1M requests and 400,000 GB-seconds per month. For a 512MB function running for 100ms, that's 0.512 GB × 0.1s = 0.0512 GB-seconds per invocation.

Workers has two plans: Free (100K requests/day, 10ms CPU/request) and Paid Bundled ($5/month for 10M requests, then $0.50/M additional, 50ms CPU). Workers Unbound is $0.015/M requests + $0.0000125/GB-second CPU time. Workers are significantly cheaper than Lambda for high-volume short requests because there's no per-GB-second charge on Bundled pricing.

Vercel Edge Functions on Pro plan ($20/month) include 1M invocations. Additional invocations cost $2/M on Pro. Edge Functions run on Vercel's edge network (powered by Cloudflare) with very low latency. They're best for auth middleware, redirects, A/B testing, and geolocation — not heavy compute.

For very high volume (100M+ requests/month), Cloudflare Workers Unbound is often cheapest, especially for CPU-light workloads. For moderate volume (1-10M requests) with mixed compute, Lambda can be competitive. Vercel Edge is competitive for Next.js apps already on Vercel. Always compare at your specific request volume and duration profile.

This calculator covers compute costs only. Not included: data transfer/egress costs (Lambda outbound: $0.09/GB), API Gateway costs ($3.50/M calls), KV/D1/R2 storage (Workers), DynamoDB/RDS (Lambda), cold start impact on user experience, monitoring/logging costs, and developer time for migration.