All GPT integrations I’ve implemented have been via Azure’s service, due to Microsoft’s contractual obligation for them not to train on my data.
As far as I understand it, this ruling does not apply to Microsoft, does it?
I think when you spin up open AI in azure, that instance is yours, so I don't believe that would be subject to this order.
The plans scale down far enough that they can't possibly cover the cost of a private model-loaded-to-vram instance at the low end.