The need of a device providing two translational (2T) and two rotational (2R) movements led us to the design a 3UPS-1RPU parallel manipulator. The manipulator consisted on a mobile platform connected to a base through four legs. That is, the manipulator layout has one central leg and three external legs at the same radial distance. By studying different locations of the legs anchoring point, we improved the first layout design, yet not the optimal one. On this basis, this paper focus on the optimal dimensional design of the manipulator. To this end, we put forward the kinematics equations of the manipulator in accordance to the anchoring points coordinates. Through a numerical approach, the equations enable to find the manipulator workspace. Also, we find a global manipulability index using a local dexterity measure. The latter index serves as optimal function. The optimization process considers joint constraints. Thus, we built a nonlinear optimization problem solved through sequential quadratic programming algorithms. We start by optimizing only a small set of parameters rather than the entire set, which gives us insights on the initial guess to optimize using the entire set. The optimal design layout varies from the original layout. Findings suggest that a task-oriented reconfiguration strategy can improve manipulator performance.