This paper proposes a method for designing the 3D deformable shape of a continuum robot based on genetic algorithm optimization for the arrangement of tendon guides. We introduce plain and penetration models that are classified based on tendon configurations. A combination of these models produces 3D deformations. Optimizing the arrangement of the tendon guides extends the behavior of the prototype by providing a distribution in its curvature. We designed a model to simulate the deformed shape of the prototype and used a genetic algorithm to determine the optimal solution for the parameters to be mathematically suited to the given shapes. Additionally, we built prototypes using the applied parameters and demonstrated them experimentally. We also verified the continuum robots with optimized parameters by comparing them with pre-optimized parameters. The results of this study can contribute to improving the dexterity and versatility of the movements of continuum and biomimetic robots using only a few actuators.