Actually, as a person who rented a dedicated server or two (not from aws) this exact thing happens quite often.

For example a provider quietly underclocks your cpu or puts it is an energy saving plan hoping you won't notice.

I think they could easily reduce several parameters to slightly reduce cost (something like "it's the same model, a few less passes, but quality is still 95%+").

Or switch to a faster, cheaper cache mechanism or something that's almost as good, but much cheaper to run, in the name of improving efficiency and keeping up with demand. This switch to cheaper mechanisms is what it feels like to me.

It's a problem for us. When they nerf the model our productivity really drops (due to the specific use case and process we have). We've started using OpenAI and will gradually use it more and more if this continues.