This work proposes multi-modal variants of the well-known Vehicle Routing Problem. It will be empirically shown that even a relatively simple metaheuristic (namely, the Large Neighborhood Search) is able to highlight the potential benefit of multi-modality.