Monitoring partial vapor pressure in the freeze-drying chamber is a cheap, global, and non-intrusive way to assess the end of the primary drying stage. Most existing dynamic freeze-drying models which predict this partial pressure describe mass transfer between the product and the condenser via a mass transfer resistance or a mass transfer coefficient. Experimental evidence suggests that such models can be significantly in error for some values of the sublimation flux, leading to physically inconsistent predictions and possibly incorrect assessment of primary drying termination, with potential risk of product damage if moving to secondary drying and increasing shelf temperature while some ice is still present. Assuming a binary gas transport model for vapor and inert gas leads to improved and consistent predictions and explains the apparent variation of the mass transfer resistance with total pressure, shelf temperature, and product sublimation area.