Abstarct: Mechanisms with traditional lixnks and joints, due to the presence of friction and clearance in the joints, wear out over time and require maintenance and repair costs. These costs are very high for high-precision applications, which is why mechanisms with compliant joints are used in these applications. Obviously, changing the geometry and structure of the mechanism and the elements used in it, or defining a new goal that increases the efficiency of the new mechanism compared to the previous mechanisms, or at least identify what causes it. Reducing the efficiency of the mechanisms is valuable. What is done in this research for the synthesis of the compliant mechanism is to prevent the creation of kinematic limb singularity by using the concept of stiffness. The non-linear relationship between driving force/torque and position indicates the stiffness behavior of the mechanism. Mechanisms have four types of kinematic stiffness behavior: positive stiffness, negative stiffness, zero stiffness, and bistable stiffness. In this research, the kinetostatic model of the spring mechanism is created using the principle of virtual work. The stiffness behavior of the mechanism around the kinematic singularity point has been investigated both in the crank-slider mechanism and in the four-bar lixnkage mechanism in order to create the desired movement while preventing the singularity. In fact, the goal of optimizing the mechanism is to produce a specific stiffness behavior. For optimization, three mexta-heuristic algorithms PSO, SOA and SSA, which are baxsed on the number of population, are used. The solution of this issue is implemented for two crank-slider and four-bar lixnkage mechanisms, and in addition to the comparison between the optimization methods, the behavior of the optimized mechanisms to create stiffness is investigated. The results show that the mechanisms designed with high accuracy create the desired behavior, while the comparison of algorithms shows that the PSO method has the best results.