Injecting CO2 into the reservoir has the dual potential to enhance oil recovery and sequestrate CO2 underground, thereby reducing the CO2 level in the atmosphere. In the process of CO2 flooding, the recovery of CO2 miscible flooding is higher than that of CO2 immiscible flooding. Determining the minimum miscibility pressure (MMP) is therefore crucial for CO2 flooding. Experimental methods can accurately determine the MMP of CO2 and crude oil, but are time-consuming and costly. Empirical correlation-based methods can rapidly calculate MMP CO2 and crude oil, but their applicability is limited due to the small number of CO2-oil samples considered. This paper proposes an ensemble model based on the integration of multiple machine learning algorithms to predict the MMP between pure CO2 and crude oil. The method collects a dataset of 165 experimental MMP data on pure CO2 miscible flooding and classifies influencing factors into four attributes: reservoir temperature, the molecular weight of C5+ oil fraction, and the percentage of volatile and intermediate components in the crude oil. These attributes are used as inputs to train the ensemble model. By comparing the predicted MMP results of the proposed model with those obtained from other machine learning methods, empirical correlations, and experimental results, it can be concluded that the proposed ensemble model is more accurate. The impact of data attributes on the proposed ensemble model is verified through univariate analysis, which provides evidence that the proposed method aligns with physical cognition. Therefore, the methodology presented in this paper can serve as a reference for forecasting MMP of CO2 and crude oil.