In the realm of engineering, smart multifunctional composites (SMFCs), such as electroactive polymers (EAPs), magnetoactive polymers (MAPs), thermoactive polymers (TAPs) and functional hydrogels, that can be programmed to perform a variety of tasks by intelligently exploiting bending, twisting and/or buckling through the application of external stimuli. Simulation of SMFCs require accurate resolution of large-deformation and large-strain behaviour coupled with the field variables and their derivatives. Although several finite element formulations have been proposed for the coupled simulation of SMFCs, majority of these formulations use monolithic approaches where fields are solved in a single big matrix [1], [2]. While partitioned schemes have been explored to a limited extent [3], [4], the existing partitioned schemes struggle to maintain stability, accuracy and computational efficiency when the severity of coupling increases, for higher values of applied fields resulting in larger deformations and strains. This work proposes partitioned schemes for SMFCs with improved stability properties that can handle higher nonlinearities in coupling. The proposed schemes employ a predictor-correct approach with predictors that substantially improve stability of partitioned schemes. The results of numerical examples show that proposed schemes are quite promising in handling higher nonlinearities in coupling. The accuracy and stability of the proposed partitioned schemes is demonstrated by studying a few benchmark examples over a range of applied loads. The computational efficiency of the proposed schemes is demonstrated by studying the examples with different mesh sizes and by comparing against the computational cost of monolithic schemes.