Abstract We propose an effective model for solute separation from fluids through reverse osmosis based on core-softened potentials. Such potentials have been used to investigate anomalous fluids in several situations under a great variety of approaches. Due to their simplicity, computational simulations become faster and mathematical treatments are possible. Our model aims to mimic water desalination through nano-membranes through reverse osmosis, for which we have found reasonable qualitative results when confronted against all-atoms simulations found in the literature. The purpose of this work is not to replace any fully atomistic simulation at this stage, but instead to pave the first steps towards coarse-grained models for water desalination processes. This may help to approach problems in larger scales, in size and time, and perhaps make analytical theories more viable.