There are a plethora of models available for simulation of river flows, ranging from conceptual to physically based hydrological models, land surface models, and increasingly incorporating machine learning approaches. Given this diversity of models, it is vital that we implement systematic model intercomparisons, multi-model approaches and model benchmarking to learn about the relative merits of different model structures, guide model development, and quantify model structural uncertainties. This session aims to bring together researchers and practitioners involved in river flow modelling, with a focus on multi-model approaches and methods to evaluate and learn from model differences.