No? The api has all the models

GPT-4 API is very expensive, there is no way for someone to use it intensively every day and end up with a $2 bill. With GPT-3.5 it's normal.

I must be hallucinating that bill then.

I am surprised by the low bill too. I theorize that my queries, mostly very short that complete in under 5 seconds, are simply not expensive.

GPT4 use will cost you a good deal more than GPT3.5Turbo.