Realtime API / Tokens per second?

Hello
I am trying to find information/data about the number of toekns per second delivered for each model, in order to get some performance figures.
Thanks for your insight
FQ