Running a prompt in the cloud vs. running a prompt locally on a laptop (even when the models are different) does give a rough sense of the resources that are being used in the cloud: