This paper aimed to establish optimality conditions and develop an efficient solution method for unconstrained fuzzy optimization problems using the steepest descent method, based on the granular differentiability and granular convexity of fuzzy mappings. First, building upon the gr-difference for fuzzy numbers, the granular gradient, gr-differentiable and twice gr-differentiable of fuzzy mappings were researched. Furthermore, the efficient solution of fuzzy optimization was introduced under the granular convexity of fuzzy mappings. Finally, we proposed a steepest descent method for fuzzy optimization problems by employing the characterization function of fuzzy mappings. Our analysis demonstrated that the proposed method converged linearly under standard convexity conditions.