The practical implication of nanofluids is essentially dependent on their accurate modelling, particularly in comparison with the high cost of experimental investigations, yet the accuracy of different computational approaches to simulate nanofluids remains controversial to this day. Therefore, the present study is aimed at analysing the homogenous, multiphase Eulerian–Eulerian (volume of fluid, mixture, Eulerian) and Lagrangian–Eulerian approximation of nanofluids containing nonspherical nanoparticles. The heat transfer and pressure drop characteristics of the multiwalled carbon nanotubes (MWCNT)-based and multiwalled carbon nanotubes/graphene nanoplatelets (MWCNT/GNP)-based nanofluids are computed by incorporating the influence of several physical mechanisms, including interfacial nanolayering. The accuracy of tested computational approaches is evaluated by considering particle concentration and Reynolds number ranges of 0.075–0.25 wt% and 200–470, respectively. The results demonstrate that for all nanofluid combinations and operational conditions, the Lagrangian–Eulerian approximation provides the most accurate convective heat transfer coefficient values with a maximum deviation of 5.34% for 0.25 wt% of MWCNT–water nanofluid at the largest Reynolds number, while single-phase and Eulerian–Eulerian multiphase models accurately estimate the thermal fields of the diluted nanofluids at low Reynolds numbers, but overestimate the results for denser nanofluids at high Reynolds numbers.