This paper presents the development of accurate map-based models for characterizing a new Dual Source Heat Pump prototype. This unit includes three braze plate heat exchangers and a round tube fin heat exchanger, allowing the unit to select different operating modes such as heat pump, chiller, and domestic hot water production using as source the ground or air. Therefore, due to the hybrid typology of this unit and the possibility of reversing the cycle, this work covers the main heat pump and refrigeration equipment technologies currently available on the market (air source and ground source units). The modeling strategy selected has been to provide several polynomial expressions to predict the performance of these units, i.e., compressor energy consumption and condenser and evaporator capacities. This approach allows obtaining accurate, compact, and easy-to-implement models for developing dynamic models of more complex systems where this type of unit – the heat pump – is an integrated part of the system. Currently, a clear example of this modeling strategy can be found in characterizing one of the main components installed in these machines, the compressor. The AHRI-540 standard specifies a polynomial model as a function of evaporating and condensing temperatures. In this sense, for the characterization of heat pumps, the polynomials developed depend only on the unit’s external variables, so they can be useful in many scenarios, obtaining direct feedback on the heat pump performance when developing a dynamic model to optimize system control strategies or to develop techno-economic studies. In this case, the hybrid typology of this unit makes it particularly relevant to optimize the control to manage the type of source to be used (air or ground), allowing the development of a more efficient and sustainable technology by selecting the most adequate source in terms of performance. This study focuses on obtaining the polynomial expressions that minimize the number of terms while simultaneously minimizing prediction error. By carefully selecting the most significant terms and suitable transformations in the characterized variables, the goal is to prevent overfitting, minimize potential extrapolation or interpolation errors and obtain polynomial expressions that can be fitted with small experimental samples. For this purpose, a detailed model implemented in a commercial software for heat pump characterization has been used, with which a large number of simulation results were generated. These simulation results include a fine meshing working map of the unit that allowed us to analyze the relationships between the characterized and external variables.