Pricing · Functions

Functions credit pricing

Functions pricing on KERNSTACK is one multiplication: GiB-seconds of actual execution times 0.01 credits, drawn from the project’s single balance. This page works the arithmetic for three realistic API shapes so the number can be checked before the first deploy, and states what the hyperscaler bill would itemise instead.

How Functions is metered on one credit balance

Base Rate Summary: Compute is 0.01 credits per GiB-second of execution. An idle function costs nothing; there is no per-request fee and no minimum.

How to estimate a function workload

Two numbers predict the bill: requests per month and average GiB-seconds per request (memory footprint times seconds of execution). A small JSON handler typically lands between 0.05 and 0.3 GiB-seconds. Multiply requests by GiB-seconds by 0.01 and the result is credits; at the Builder bundle (€200 for 20,000 included credits) a credit is roughly a cent of included capacity.

What is not charged

No per-request fee, no charge for declared-but-uncalled functions, no standing replica. A project with forty functions and four callers pays for four callers’ worth of execution. Gateway routing itself is not a metered product line; egress is the one companion charge, at 9.00 credits per GiB when responses actually leave.

  • Preview and agent demo functions spend from the same balance only when invoked
  • Scheduled functions (cron entries) meter exactly like request-driven ones
  • The plan Calculator on this page lets you drag the two input numbers yourself

Credit Consumption Scenarios vs Hyperscaler Billing

Workload Scenario Monthly Resource Footprint KERNSTACK Credits Hyperscaler / SaaS Equivalent
Light Internal API: 200k requests/month at 0.2 GiB-s average ≈400 credits Request fee plus duration metering, a separate gateway per-request charge, and log ingestion SKUs
Steady Product API: 5M requests/month at 0.25 GiB-s average ≈12,500 credits Same stack of meters at volume; provisioned-concurrency charges appear the moment latency matters
Heavy Busy API: 60M requests/month at 0.4 GiB-s average ≈240,000 credits Duration and request meters dominate; egress billed separately per GiB on top

Frequently asked questions

No. Only execution counts: GiB-seconds at 0.01 credits. A million tiny requests and a thousand heavy ones are priced by the work inside them, which is the honest version of the same total.

Nothing. Forty declared functions with no traffic cost exactly the storage of their code — in practice, zero visible lines on the balance. Scale to zero is the default physics, not a tier.

The unit is deliberately familiar (GB-seconds) minus the surrounding meters: no separate request charge, no gateway per-call fee, no log SKU. Teams usually find the KERNSTACK line is the duration line of the old bill, alone.

Ship your stack on the Software Lifecycle Cloud.

Closed preview on dedicated European infrastructure. Bring a repository or start from a blueprint.