Serverless pricing looks complicated, but almost every platform’s bill comes down to four numbers: a per-request fee, a per-GB-second memory-time charge, a free tier, and egress. Get those four and you can compare any two platforms.
The headline rates (June 2026 snapshot)
| Platform | Per 1M requests | Per GB-second | Free tier (perpetual) |
|---|---|---|---|
| AWS Lambda | $0.20 | $0.0000167 | 1M req + 400k GB-s |
| Azure Functions | $0.20 | $0.000016 | 1M req + 400k GB-s |
| Cloudflare Workers | $0.30 | CPU-ms model | 100k req/day, no egress |
| Google Cloud Run | $0.40 | $0.0000025/GiB-s + vCPU-s | 2M req + 360k GiB-s |
| Vercel Functions | $0.60 | active-CPU + memory meters | Hobby: 1M req hard-capped |
| Scaleway | ~$0.165 | ~$0.0000055 | 1M req + 400k GB-s |
| DigitalOcean Functions | no per-req fee | $0.0000185/GiB-s | 90k GiB-s |
Snapshot captured June 2026 — verify on each vendor’s pricing page.
The per-request fee is rarely the big number
A $0.20-vs-$0.40 difference per million requests is real but small. The charge that usually dominates is GB-seconds = requests × duration × memory. Doubling your function’s memory or its run time doubles that cost. See the cheapest-per-GB-second ranking.
Egress is the silent budget-killer
Hyperscalers bill outbound data at roughly $0.09/GB after a small free allowance. Cloudflare Workers charges nothing for egress, which often makes it the cheapest end-to-end option for content-heavy APIs even though its per-request fee is higher.
Bottom line
Start on a free tier — most small apps never leave it. When you outgrow it, compare the whole bill (requests + GB-seconds + egress + any platform minimum), not just the per-request rate. Estimate yours with the cost calculator or a ready-made workload scenario.