This paper presents a validation of a computational fluid dynamics (CFD) code (CUPID) for predicting the erosion of helium stratification by a vertical steam jet. For the validation of CUPID, CFD grade experimental data obtained from the PANDA facility in the framework of the OECD/NEA HYMERES-2 project was used. To predict the stratification erosion phenomena appropriately, the turbulence model and radiative heat transfer model need to be employed as major physical models. Therefore, the standard k–ε turbulence model, including the buoyancy term, was applied, and a sensitivity study was carried out to investigate the effect of buoyancy production in the turbulence dissipation rate equation on the predicted helium concentration. Additionally, a radiative heat transfer model (P-1) was implemented by applying the appropriate parameter for the calculation, and the predicted gas temperature distributions with and without radiative heat transfer were compared. First, the physical models were validated for a baseline test case in which the vertical steam jet did not interact with any flow obstacles. Then, a second validation was carried out as a blind benchmark calculation where the experimental data were released after the calculation results had been submitted. The CUPID calculations for the helium concentration and gas temperature were very close to the measured data and showed that the analysis models can be used to accurately predict the gas mixing phenomena in the containment.