I have a GPU service that will be occasionally executed and run for like 30 seconds. From what I understand the container will continue to run for 5 minutes before going back to sleep. So I am essentially paying for 5 min 30 seconds of GPU usage when only 30 seconds was used for work. Please correct me if I’m wrong.
I also saw that the idle period can be customized on paid plans. So I wonder can it be set lower than the default 5 minutes? If so, what is the lowest idle period possible?