In this paper, we revisit the low-thrust multi-revolution orbit-raising problem formulated as a sequence of optimal control sub-problems, each of which computes the trajectory over a planning horizon. Each sub-problem solves an unconstrained optimization problem described in terms of dynamical coordinates, by minimizing a convex combination of three components reflecting the deviation of the maneuvering spacecraft from the geosynchronous equatorial orbit. This paper explores the impact of planning horizon time-length on the optimality gap of the computed solutions, by considering minimum-time solutions as the reference. Numerical results are presented by considering transfers starting from non-planar geosynchronous transfer orbits.