The development and implementation of wheel-rail contact models in multibody codes are two active research topics, aiming at improving the accuracy of numerical results and computational efficiency of the dynamics analysis. However, the realism of numerical results is challenged when considering switches and crossings (S&C), where the most adverse wheel-rail contact conditions occur. This paper presents a benchmark study where the performance of the multibody codes MUBODyn, VOCO and VI-Rail are assessed using three case scenarios that involve typical contact conditions observed in S&C. A focused description of the relevant methods to determine the wheel-rail contact forces is presented for each software. The three scenarios considered in this work have been designed to represent typical challenging contact conditions observed in S&C, i.e. conformal contact, contact with a sharp edge, and impact loads. The scenarios proposed in this work are fully described, making them easily reproducible. The agreement between results is discussed in the framework of the methods implemented in each code. This work highlights the impact of wheel-rail contact methods on the results as well as on the computational efficiency of the multibody codes.