Determining solubility of hydrocarbon and non-hydrocarbon components of natural gas is crucial for theoretical studies and engineering design. In this study, new solubility prediction models were developed for both hydrocarbon gases (methane, ethane, propane, and butane) and non-hydrocarbon gases (CO2 and N-2) in aqueous solutions of strong electrolytes using a hybrid modeling strategy, which links the Coupled Simulated Annealing (CSA) to the Least-Squares Support Vector Machine (LSSVM) technique. Comparing the models' predictions with experimentally determined solubility values, a very good agreement was noticed, leading to the overall correlation coefficients of 0.9880 and 0.9907 for the hydrocarbon and non-hydrocarbon gases, respectively. These models were also found to succeed in capturing the physical trends among experimental datasets through performing sensitivity analysis between the dependent and independent parameters. Developed models can be utilized to predict the solubility of pure and/or a mixture of hydrocarbon and non-hydrocarbon gases in aqueous electrolyte solutions, covering wide ranges of ionic strength, pressures, and temperatures up to supercritical conditions. Such a reliable predictive tool helps researchers and engineers to successfully obtain the key thermodynamic properties (e.g., solubility, vapor pressure, and compressibility factor), which are central to properly design and operate the corresponding units in a variety of chemical plants such as petrochemical plants, natural gas processing plants, and refineries.