Nowadays, buildings are responsible of a large consumption of energy in our cities. Moreover, buildings can be seen as the smallest entity of urban energy systems. On these premises, in this paper we present a flexible and distributed co-simulation platform that exploits a multi-modelling approach to simulate and evaluate energy performance in smart buildings. The developed platform exploits the Mosaik co-simulation framework and implements the Functional Mock-up Interface (FMI) standard in order to couple and synchronise heterogeneous simulators and models. The platform integrates: i) the thermal performance of the building simulated with EnergyPlus, ii) the space heating and hot water system modelled as an heat pump with PID control strategy in Modelica, and iii) different Python models used to simulate household occupancy, electrical loads, roof-top photovoltaic production and smart meters. The platform guaranties a plug-and-play integration of models and simulators, hence, one or more models can be easily replaced without affecting the whole simulation engine. Finally, we present a demonstration example to test the functionalities and capabilities of the developed platform, and discuss future developments of our framework.