AbstractPhotovoltaic systems are increasingly applied in buildings to reduce energy costs and support sustainability. Their performance depends on factors such as panel angle, orientation, and temperature, which significantly affect energy yield. This study compares the actual output of an operating photovoltaic system with results from three calculation methods: manual estimation, 3D modeling, and artificial intelligence-based evaluation. The photovoltaic geographical information system and Rhino-Grasshopper methods proved most accurate, deviating by less than 10% from measured data due to their use of extensive, long-term meteorological datasets that ensure reliable performance prediction.