The accuracy of the modelling of gas–liquid flows depends strongly on a suitable modelling of the interfacial forces. Among these, drag is dominant. Most drag models reported in the literature have been derived and validated only for laminar or low-turbulent flow conditions. In this study, we numerically evaluated several drag models from the literature for high-turbulent gas–liquid flow around an obstacle in a pipe that creates a distinct vortex region. We performed Computational Fluid Dynamics (CFD) simulations and compared the void fraction and gas velocity profiles with experimental data obtained by ultrafast X-ray computed tomography. We found that all models, except Schiller&Naumann and Feng, predicted the void fraction well compared to experimental data upstream of the obstacle, i.e., for a developed two-phase pipe flow with axial symmetry. However, the void fraction downstream is greatly overestimated by all models except those that appropriately consider the turbulence effects. Based on the results, a hybrid drag model is proposed that significantly improves the prediction of the void fraction.