Liquid metal cooled fast reactors, such as the European Sodium Fast Reactor (ESFR), require accurate predictions of turbulent heat and momentum transfer to ensure safe and efficient operation. However, the low-Prandtl-number nature of liquid metals challenges conventional thermal turbulence models, which are typically developed and calibrated for air or water flows. This limitation can lead to significant uncertainties in thermal–hydraulic predictions and, consequently, in reactor design margins. The objective of this work is to implement and validate turbulence models tailored for liquid metals within the open-source Computational Fluid Dynamics (CFD) C++ library OpenFOAM and to exploit them in a coupled framework with the French system thermal–hydraulics (SYS-TH) code CATHARE2 to assess their impact on the ESFR reactor design. An isotropic dynamic and thermal four-factor turbulence model is implemented and validated against reference experimental benchmarks for liquid metal flows, covering both forced and mixed convection regimes. The validated model is then applied to a representative ESFR hot-pool plenum, where thermal stratification may occur, and compared with native OpenFOAM turbulence models commonly used in nuclear reactor simulations. The coupled framework replaces the SYS-TH representation of the hot pool with a three-dimensional CFD model to better capture the underlying physical phenomena and quantify differences between the two approaches. The coupling strategy, based on data exchange through the open-source MED/MEDCoupling libraries, transfers the time evolution of selected quantities at the hot-pool inlet from CATHARE2 and uses them as time-dependent boundary conditions for the CFD simulations. Conversely, the outlet fields of the OpenFOAM domain are spatially averaged into representative scalar quantities and imposed as boundary conditions at the SYS-TH hot-pool outlet. This methodology is applied to both normal operating conditions and beyond-design-basis accident scenarios. The results highlight the importance of dedicated liquid metal turbulence models and of accounting for multi-scale interactions through CFD–SYS-TH coupling in nuclear reactor simulations.