This nlab page shows a chart describing the structure of the classifying topos of a theory in various logics, but it doesn't have an entry for the classifying topos of a classical first-order theory.
Do they always exist? If not, is there a standard counterexample and characterization for when it does?