Ah, did I misunderstand this para from the Cloud Run GPU docs? I can set min instances to 0 and only pay for request-triggered GPU time?
“There are no per request fees. Because you must use CPU always allocated to use the GPU feature, minimum instances are charged at the full rate even when idle.”
November 25, 2024 at 12:11 AM
Ah, did I misunderstand this para from the Cloud Run GPU docs? I can set min instances to 0 and only pay for request-triggered GPU time?
“There are no per request fees. Because you must use CPU always allocated to use the GPU feature, minimum instances are charged at the full rate even when idle.”