$ZFC$ proves that $\mathbb N$ is a model of $PA$. Even $ZF$ does. How weak can go?
In particular, is there some weak set theory that proves that $\mathbb N$ is a model of $PA$, but does not proof that $PA + Con(PA)$ is consistient?
EDIT: The theory should prove that $\mathbb N$ (along with its operations) actually exist.