We propose a model based on non-equilibrium lattice fluid (NELF) theory and corrected fractional free volume of polymers to effectively and accurately predict the solubility of gases in different polymers. The method to achieve this purpose is based on the utilization of NELF model infinite dilution solubility coefficient (S0) as the base of predictive calculations. To account for the isolated pore in the polymer matrix in density estimation, a fractional free volume correction factor (β) was introduced in NELF model. The modified NELF model was successfully applied for prediction of solubility of C3H8 and CO2 in polyethylene oxide (PEO) and CO2 in polyethylene terephthalate (PET), isotactic polypropylene (i-PP), polyetherimide (PEI), polymethyl methacrylate (PMMA) and polyethyl methacrylate (PEMA) with adjustments in β value and depth of diffusion of gases in polymer matrix (ζ) at different pressures and temperatures. This work involves multi-objective optimization using genetic algorithm of MATLAB toolbox with adjusted settings. It applies to find the optimum temperature at which the minimum standard deviation of β for different gas-polymer systems is obtained. β showed the same trend of change with temperature as the constrained pressure imposed on the amorphous phase in semi-crystalline polymers. A cubic correlation for standard deviation for β versus temperature was obtained which was able to anticipate the changing trend of β at different temperatures. The chisquare test results verified that compared with original NELF model, a more accurate model for prediction of gas solubilities in polymers has been proposed.