Machine learning assisted optimal power flow (OPF) aims to reduce the computational complexity of these non-linear and non-convex constrained optimisation problems by consigning expensive (online) optimisation to offline training. The majority of work in this area typically employs fully-connected neural networks (FCNN). However, recently convolutional (CNN) and graph (GNN) neural networks have been also investigated, in effort to exploit topological information within the power grid. Although promising results have been obtained, there lacks a systematic comparison between these architectures throughout literature. Accordingly, we assess the performance of a variety of FCNN, CNN and GNN models for two fundamental approaches to machine learning assisted OPF: regression (predicting optimal generator set-points) and classification (predicting the active set of constraints). For several synthetic grids with interconnected utilities, we show that locality properties between feature and target variables are scarce, hence find limited merit of harnessing topological information in NN models for this set of problems.