This paper presents a semi-empirical modeling approach that involves the use of physical representations and parameter tuning to enable predictions of steady-state performance for variable refrigerant flow heat pumps systems that incorporate multiple indoor units having variable-speed fans. The system model is based on the use of semi-empirical component models where parameters are tuned using data. The semi-empirical models are built based on first principles but are simplified to speed up the computation. Deviations of the model from experimental data due to model simplification are corrected by tuning the heat transfer coefficients and charge at the system level with the experimental data. Control algorithms for the electronic expansion valve are also estimated based on experimental observation. The model is built in modules to allow the flexibility to add components such as more indoor units or to change the control algorithms easily without a significant change in the solver. A case study was carried out using data from a ductless heat pump system having two indoor units and results show that the model can accurately simulate performance at conditions not used for tuning.