To enhance the working accuracy of the manipulator, this paper presents a research and optimization method for the stiffness performance of a redundant degree of freedom quasi-series manipulator. Firstly, based on the self-motion characteristics of redundant manipulator, the inverse kinematics is solved by geometric analysis method of position and posture separation. Then, a stiffness modeling method of multi-branch closed-loop mechanism based on virtual work principle is introduced, and the overall stiffness model is established. Based on the above analysis, a comprehensive stiffness performance evaluation index derived from Lagrange function is proposed to evaluate stiffness performance. Finally, the genetic algorithm is employed to optimize the stiffness performance, followed by Ansys simulation and experimental verification. The results demonstrate the effectiveness of the stiffness optimization method, yielding a remarkable 117% improvement in manipulator stiffness performance at the same position. The research results of this paper enrich the optimization methods of stiffness performance and lay a foundation for further stiffness optimization and mechanism optimization.