The accurate prediction of the minimum miscibility pressure (MMP) for sour natural gas—reservoir oil is of paramount importance for the design and optimization of gas injection processes, particularly in enhanced oil recovery (EOR) and gas cycling schemes. This study introduces intelligent models that leverages machine learning algorithms to predict the MMP between reservoir oil and injected sour natural gas. The model is trained and validated using a comprehensive dataset encompassing various oil properties, reservoir temperature, and gas compositions commonly encountered in sour natural gas. The different machine learning methods are chosen to construct the MMP forecasting model with influential parameters affecting MMP, followed by the comparison of predictive accuracy of different approaches. Validation results demonstrate that the intelligent model achieves great predicting effects. The superior performance of model is attributed to its ability to capture intricate patterns and interactions within the dataset that are often overlooked by conventional methods. Furthermore, the intelligent model offers a user-friendly interface for rapid MMP prediction, enabling petroleum engineers to make informed decisions regarding gas injection strategies without the need for extensive laboratory experiments or complex simulations. This not only enhances operational efficiency but also contributes to cost savings and risk reduction in the development under the sour natural gas injection condition. In conclusion, the integration of machine learning techniques and comprehensive datasets provides a robust and accurate tool for the petroleum industry, facilitating the optimization of gas injection processes and thus enhancement of oil recovery from challenging reservoirs.