Cloud Build Cost Calculator
Cost Cloud Build minutes by machine type after cache savings and the free tier, and check concurrency queueing.
Inputs
Layer caching and a warm dependency cache typically remove a third of a build.
A bigger machine costs more per minute but finishes sooner — the product decides the bill.
Monthly Cost
$4.79
Billed Build-Minutes
1,595minutes
Build-Minutes Consumed
4,095minutes
Minutes Saved by Caching
2,205minutes
Rate per Build-Minute
$0.0030
Cost per Build
$0.0053
Expected Queue Wait at Peak
9.8minutes
Concurrency Verdict
Comfortable — builds start almost immediately at the stated peak
Step by step
Values used
Builds per month = 900 builds; Average build time = 7 minutes; Build machine type = e2-standard-2 (default); Price per build-minute on the default machine = 0.0030 USD/minute; Price per build-minute on e2-highcpu-8 = 0.0160 USD/minute; Price per build-minute on e2-highcpu-32 = 0.0640 USD/minute; Free build-minutes per month = 2,500 minutes; Build time removed by caching = 35 %; Speed-up on the larger machine = 2.50 x; Concurrent build slots = 10 builds; Peak builds submitted per hour = 90 builds
Cloud Build Cost
billed minutes = builds × (build time ÷ machine speed-up) × (1 − cache saving) − free minutes; cost = billed minutes × the machine's per-minute rate.
Monthly Cost
= 4.79
Billed Build-Minutes
= 1,595 minutes
Build-Minutes Consumed
= 4,095 minutes
Minutes Saved by Caching
= 2,205 minutes
Rate per Build-Minute
= 0.0030
Cost per Build
= 0.0053
How it works
A faster build machine costs more per minute but runs for fewer minutes, so the bill is the product of the two — an e2-highcpu-8 at roughly five times the rate is only worse if it delivers less than a five-fold speed-up. Caching cuts minutes directly, and the free monthly allotment is subtracted afterwards, which is why small teams often pay nothing at all. Build queueing wastes engineer time rather than money, and the queue model here shows how close the concurrency limit is to saturation before developers start complaining. Confirm the per-minute rate for each machine type in the Google Cloud pricing calculator, and remember the free allotment is per billing account.
Formula
Cloud Build Cost
billed minutes = builds × (build time ÷ machine speed-up) × (1 − cache saving) − free minutes; cost = billed minutes × the machine's per-minute rate.
- speed-up
- How much faster the larger machine completes the same build
- cache saving
- Share of build time removed by layer and dependency caching
- load factor
- Peak submission rate divided by what the concurrency limit can clear per hour
Frequently Asked Questions
How is Cloud Build Cost calculated?
billed minutes = builds × (build time ÷ machine speed-up) × (1 − cache saving) − free minutes; cost = billed minutes × the machine's per-minute rate. A faster build machine costs more per minute but runs for fewer minutes, so the bill is the product of the two — an e2-highcpu-8 at roughly five times the rate is only worse if it delivers less than a five-fold speed-up. Caching cuts minutes directly, and the free monthly allotment is subtracted afterwards, which is why small teams often pay nothing at all.
Why does Cloud Build Cost matter?
Build queueing wastes engineer time rather than money, and the queue model here shows how close the concurrency limit is to saturation before developers start complaining. Confirm the per-minute rate for each machine type in the Google Cloud pricing calculator, and remember the free allotment is per billing account.
What values do I need to enter?
This calculator takes 11 inputs: Builds per month, Average build time, Build machine type, Price per build-minute on the default machine, Price per build-minute on e2-highcpu-8, Price per build-minute on e2-highcpu-32, Free build-minutes per month, Build time removed by caching, Speed-up on the larger machine, Concurrent build slots, Peak builds submitted per hour. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.
Is the biggest build machine ever worth it?
When the build genuinely parallelises — large compilations, big test suites — a 32-vCPU machine can pay for itself and return engineer time as well. For a build that spends most of its time pulling dependencies over the network, the extra vCPUs sit idle and you pay the higher rate for nothing.
You might also need
- Cloud Run Cost CalculatorCommonly used together
- Artifact Registry Storage CalculatorCommonly used together
- Container Registry Migration CalculatorCommonly used together
- BigQuery Cost CalculatorAlso in Google Cloud
- Compute Engine Cost CalculatorAlso in Google Cloud
- Cloud Functions Cost CalculatorAlso in Google Cloud