Am I understanding correctly that the Start OS app is for processing the request on one's personal node hardware?
If so, can one select how much computer resources to allocate?
I am imagining all the poor little Raspberry PIs that will be melting down under the load.
It allows you to sell ChatGPT proxy services to the network for sats.
You just proxy requests (as opposed to running the computationally expensive inference algorithm to generate content) so the additional load caused by CASCDR is pretty minimal. The real issue would be running the lightning node in the first place on a Raspberry Pi which is a risky proposition.
I would strongly advise against running a node on a RPi in the first place and I would say running FreeGPT would be completely out of the question. I highly recommend S9's Intel NUC.