1
Ask HN: Cheaper way to do model inference?
Does anyone know of any solutions for saving GPU compute during server downtime? Is there a managed solution to turn off a pod and turn it back on when I need it? I'm currently doing model inference and most of the time I'm just paying for compute without serving any user requests.