Equilibrium climate sensitivity (ECS) and its related feedbacks are important metrics used to measure the global mean surface temperature change in future climate projections. This paper uses the radiative kernel approach and a simplified cloud feedback calculation (comparing three different cloud feedback methods) to analyze the differences in the ECS, as well as the feedbacks contributing to it, between two versions of the Flexible Global Ocean-Atmosphere-Land System model (i.e., FGOALS-g2 and FGOALS-g3). The results show that the ECS of FGOALS-g3 is smaller than that of FGOALS-g2 (2.8 K versus 3.3 K). The main feedbacks contributing to the ECS change in FGOALS-g3 are the weaker surface albedo feedback and stronger negative shortwave cloud feedback. The reduced surface albedo feedback in FGOALS-g3 is associated mainly with its mean base state, which has a lower surface air temperature and larger sea ice area compared with FGOALS-g2. The enhanced negative shortwave cloud feedback in FGOALS-g3 is caused mainly by the larger low-cloud area fraction and liquid water path. Furthermore, the ECS change can be traced back to the different cloud parameterization scheme, parameter tuning, ocean grid, and external forcings used in FGOALS-g3, as these all affect the mean climate state of the model.