Comment by okpatil

Atom is 60M Param (around 133x to 400x smaller).

16ms latency. And locally run.

https://at0m.pienomial.com/

Why go big when you can go small ?