Turbulence modeling is a key feature of CFD design tools, which improvement directly impacts the design margin of aeronautical products. Second-moment Reynolds-Averaged Navier-Stokes (RANS) models, also called Reynolds Stress Models (RSM), reduce the empiricism of the Eddy Viscosity Models (EVM) and render inherently the anisotropy of turbulence for 3D and wall-bounded cases, and are therefore well-adapted to the 3D highly anisotropic flows encountered in the aeronautical applications. Their spreading in the industry has mainly suffered for a long time from numerical difficulties and a higher computational cost. Recent models have been developed with emphasis on their robustness in order to be implemented in industrial CFD codes: the elliptic blending Reynolds Stress Model (EB-RSM) [1], the SSG/LRR-ω model by the DLR [2] and the SSG-ω model by the ONERA [3]. In the present work, these RSM will be benchmarked with EVM on cases of industrial interests, including the Common Research Model used in the AIAA Drag Prediction Workshop [4] and a generic Falcon configuration. The influence of their features will be investigated, such as their near-wall models (wall-distance free elliptic blending approach vs damping functions) and the influence of their chosen length-scale. The EB-RSM uses the dissipation rate ε while the others are based on the specific dissipation rate ω, which offers better predictions in adverse pressure gradient flows but degraded boundary conditions. Preliminary results show improvement of the RSM over EVM in the pressure coefficient prediction on the CRM case, especially after the shock on the suction side of the wing, confirming the robustness of RSM and their promising benefits to aeronautical cases.