# If a statement holds for all standard models of PA, then does it hold for all models?

Suppose that $\varphi$ is a consequence of every standard model of PA. Then is it provable from PA?

-
Standard examples of sentences $\varphi$ showing that this is not the case: $\mathrm{Con}(\mathsf{PA})$, "Hercules wins the Hercules-Hydra game", $\mathrm{Con}(\mathsf{ZFC}+$"there is a measurable cardinal"$)$, etc. –  Andres Caicedo Jul 5 '14 at 3:16
More interesting is to ask whether if a statement holds in all nonstandard models of $\mathsf{PA}$, then it also holds in the standard model. (And yes, that's the case.) –  Andres Caicedo Jul 5 '14 at 3:17
@Andres: (... trivially, because thanks to the upward Löwenheim-Skolem theorem some of the nonstandard models are elementarily equivalent to the standard one) :P –  Henning Makholm Jul 5 '14 at 14:03