The selection of appropriate ground motion models (GMMs) for seismic hazard analysis is of paramount importance. In recent years, there has been a significant increase in the number of ground motion models (GMMs) at both global and local levels. These models employ seismic parameters to predict ground acceleration resulting from earthquakes within a specific region. Despite this diversity, selecting the appropriate model for a specific region has always been challenging. The selection and weighting of ground motion models (GMMs) is a complex process that has recently attracted the attention of researchers. In this study, 346 accelerograms until 2025 were utilized to rank various models and accurately predict the peak ground acceleration (PGA) in one of Iran's most seismically active regions (Zagros seismotectonic province). These accelerograms are from earthquakes with a moment magnitude (Mw) of 5.0 or greater, located within a 200 km radius of the earthquake epicenter, and for which the shear wave velocity of the site was pre-existing. In this paper, eight attenuation relationships from Saffari et al.(2012,2018), Zafarani et al.(2024) and Farajpour et al.(2019) at the local level, Campbell and Bozorgnia (2014), Boore et al. (2014), Abrahamson and Silva (2014), Chiou and Youngs(2014) and Idriss(2014) at the global level were selected and Ultimately, the performance of each model in accurately predicting peak ground acceleration (PGA) was evaluated using the log-likelihood test (LLH) and compared with observed values for the Zagros seismotectonic province. The correct selection of attenuation relationships and logical weighting coefficients for them leads to a more accurate assessment of the ground acceleration and a more precise design.