Non-Gaussian states are challenging to prepare but can offer an intrinsic advantage in terms of ther metrological power. The question of whehter such states offer a true metrological advantage compared with their Gaussian counterparts when e.g. the time for state-preparation is taken into account is still largely an open question. In this talk, I will give an overview of these results and share a recent investigation into the preparation of hightly non-classical two-mode binomial cat-states.