The mixed-flow pump performs well in transporting gas–liquid mixtures with large flow rates and high inlet gas volume fractions. However, its extensive development is still limited by the operating range and poor overall performance. In this study, a multiobjective and multicondition optimization methodology for improving the gas–liquid flow performance of a mixed-flow pump based on the inverse design is proposed. The impeller blade load is taken as the optimization variable. Moreover, the Euclidean distances of the pressure increment, efficiency, and gas volume fraction in the diffuser are adopted as optimization objectives. Results show that the numerical methodology is verified by the pressure increment and gas distribution obtained in the tests of the original pump. Optimized results demonstrate that the slope of the straight line, the stacking angle at the hub, and the intersection point of the rear parabola and the straight line significantly affect the objectives. The increased load of the first half of the impeller and the reduced load of the second half may improve the comprehensive performance of mixed-flow pumps. The blade length and wrap angle are reduced after optimization, which changes the inlet and outlet angles and deflection of the blade, thereby helping to enhance the cognition of mixed-flow pump performance optimization.