Practical guide
Estimate Cloud Run cost from instance time
Calculate Cloud Run request-based or instance-based CPU, memory, requests, and minimum-instance cost.
Updated · Sources checked
Model billable CPU and memory time
Cloud Run bills CPU and memory according to its selected billing setting. Google says request-based services bill during startup, shutdown, and request handling; instance-based billing covers an instance lifetime. Minimum instances can introduce idle charges. Requests can also be a separate component.
For rate-free math, 1 vCPU and 0.5 GiB used for 100,000 billable seconds produces 100,000 vCPU-seconds and 50,000 GiB-seconds. Add request count, then subtract only the documented free-tier quantities that apply. Round according to Google’s current billing rules.
Use actual billable-instance-time metrics after deployment; latency alone is not enough. The Cloud Run Cost Calculator models request-based active CPU, memory, and requests in Iowa after its published free tiers; it excludes idle minimum-instance pricing.
FAQ
Does scale to zero cost nothing?
Without minimum instances, an unused request-based service can avoid active resource time, but configured supporting services can still cost money.