What running containers costs by vCPU, memory and hours.
Unlike functions, containers do not scale to zero — 730 hours a month regardless of traffic. Reserve capacity for steady load. Compare spiky workloads on the serverless cost calculator.
The Container Hosting Cost Calculator estimates your monthly bill for running containerized workloads based on the resources each task consumes. It combines your number of running tasks, the vCPU per task, and the memory (GB) per task with your provider's price per vCPU-hour and price per GB-hour, then multiplies by the hours per month each task runs. The result reflects the two main cost drivers of most container platforms: allocated compute and allocated memory, billed by the second or hour rather than by actual usage.
The key trade-off to watch is over-provisioning. Because you pay for reserved vCPU and memory whether or not a task fully uses them, oversized tasks quietly inflate the bill, and the effect multiplies across every running task and every hour. Before scaling up task count, right-size vCPU and memory to real observed load, and remember that tasks running 24/7 (roughly 730 hours a month) cost far more than intermittent or scheduled ones.
Serverless container platforms bill per vCPU-hour and GB-hour of allocated resources, times the hours running. An always-on service pays for all 730 hours in a month, so idle time is not free like it is with functions.
Functions win for spiky, short work and scale to zero. Containers win for steady load, long-running processes and predictable latency. Many apps use both — this calculator shows the container side so you can compare against function GB-seconds.