can't run on my dev pc forever

Most options (huggingface inference endpoint, modal, aws) run for about $1-ish an hour

The thing with hosting a model is that you basically need 24 hour uptime

oh i see

using your laptop or desktop as a server is the cheapest option