—best option
—per 1M requests
—cheapest vs priciest
Monthly cost by gateway
Ranked cheapest first for your volume. Request charges only — data transfer, caching and WAF are billed separately.
| Gateway | $/1M requests | Monthly total | Notes |
|---|
⚠️ Reference 2026 request-based rates. Real bills also include data transfer out, optional caching (charged per GB-hour on AWS), WAF, custom domains and, for Azure/tiered plans, fixed monthly instance fees not modelled here. The Cloudflare figure assumes a Workers-based gateway, a different product from managed API gateways. Confirm current pricing with each provider. ·
Report outdated price →
The gateway can cost more than the code behind it
Teams obsess over compute and forget the toll booth in front of it. An API gateway charges a small fee for every request that passes through — and at scale those small fees add up to a line item that occasionally exceeds the backend it protects. A million requests through AWS's full REST API costs about $3.50 in gateway fees before a single line of your function runs; if that function is a fast, thin handler, the gateway is the expensive part. The single biggest lever most teams miss is that AWS sells two gateways: the feature-rich REST API and the lean HTTP API at roughly a third of the price. If you're not using request transformation, usage plans or the other REST-only features, moving to the HTTP API can cut your gateway bill by around 70% with no change to your backend.
The honest way to reason about per-request economics is to add the gateway and the compute together. Price the gateway here, price what runs behind it on the serverless function cost calculator or your container platform, and read the combined cost per thousand requests. That total — not either half alone — is what each API call actually costs you, and it's the number to base pricing, rate limits and free-tier decisions on. For very high volume, simple routing, a Cloudflare Workers gateway often undercuts the managed gateways enough to change the architecture decision entirely.
How to use it
1. Enter your monthly request volume — total calls hitting the gateway.
2. Choose whether to apply each provider's free tier.
3. Read the ranked table. Note the gap between AWS REST and HTTP API — for many services that switch is the easiest saving available.
Common mistakes
Using REST API when HTTP API would do. The advanced features cost roughly 3× per request — only pay for them if you use them. Forgetting data transfer. Egress is billed on top of request fees and can dominate for large responses. Ignoring caching. Gateway caching cuts backend load but is itself billed per GB-hour — worth it only for genuinely repeated requests. Pricing the gateway in isolation. Always add the compute behind it for the real per-call cost.
Estimate only. Request-based rates; add data transfer and fixed tier fees for a full budget.
API Integration Cost CalculatorKubernetes Cluster Cost CalculatorVercel Cost CalculatorSupabase Cost CalculatorMessage Queue (SQS) Cost Calculator
How this calculator works
This API Gateway Cost Calculator estimates your monthly bill for running a managed API gateway based on a single input: your requests per month. It converts that volume into a projected charge across several providers side by side, including AWS API Gateway (comparing the pricier REST API against the cheaper HTTP API), Azure API Management, and Google Cloud. The main cost drivers are your total request count and each provider's per-request rate, which typically drops at higher usage tiers. An optional Apply free tiers? toggle lets you include or exclude introductory allowances so you can see both the promotional and steady-state price.
The key trade-off to watch is that request volume, not features, usually decides the winner: at low traffic a free tier can make one provider effectively zero, while at high traffic small per-request differences dominate. Toggle free tiers off to model your real long-term cost, since those allowances expire or cap out. Also remember gateway pricing is only one line item—data transfer, caching, and backend compute are billed separately and can exceed the gateway charge itself.
Frequently asked questions
How much does an API gateway cost?
API gateways are billed per request, usually per million. In 2026 AWS API Gateway charges about $3.50 per million calls for the full REST API and roughly $1.00 per million for the leaner HTTP API; Google Cloud API Gateway is around $3.00 per million; Azure API Management's consumption tier is about $3.50 per million after a free allowance. A Cloudflare Workers-based gateway is far cheaper per request but is a different product. So a service handling 50 million requests a month might pay anywhere from tens to a few hundred dollars just for the gateway, before the compute behind it.
What's the difference between AWS REST API and HTTP API pricing?
AWS API Gateway offers two products. The REST API is the full-featured one — request/response transformation, API keys, usage plans, WAF integration, caching — and costs about $3.50 per million requests. The HTTP API is a stripped-down, lower-latency gateway with the essentials, priced around $1.00 per million, roughly 70% cheaper. If you don't need the REST API's advanced features, switching to HTTP API can cut your gateway bill dramatically at scale. The calculator shows both so you can see the gap for your volume.
Is the API gateway or the compute behind it more expensive?
It depends on how heavy your backend is. For thin, fast functions the gateway's per-request fee can rival or exceed the compute cost — a million requests through AWS REST API costs $3.50 in gateway fees alone, which can be more than the Lambda time behind it. For heavy or slow backends the compute dominates. Model both together: use this tool for the gateway and the serverless function cost calculator for what runs behind it, then add them for the true per-request cost.
How can I reduce API gateway costs?
Switch from a full REST API to a lightweight HTTP API if you don't need the advanced features — that alone can cut the per-request rate by around 70%. Use gateway caching so repeated identical requests don't hit the backend or count as fresh billable work where applicable. Batch or coalesce chatty client requests to reduce call volume. And for high-volume, simple routing, a Cloudflare Workers gateway can undercut the hyperscaler gateways substantially, though it's a different feature set.