This chapter explores the application of various metaheuristic optimization algorithms to improve the clustering process, addressing the challenge of finding optimal centroids for the clusters. We present an in-depth analysis of the problem and how to tune the algorithms. In addition, we compare 10 metaheuristic algorithms using a MATLAB implementation on 18 different cluster tests distributed over 3 other datasets with different variations in their data sets with different variations in their functions. The chapter investigates the convergence behavior of these algorithms when considering diverse databases, which include Shapes, Synthetic, and UCI datasets. Emphasis is placed on selecting algorithms based on their performance regarding computational speed and function access. The iterative nature of clustering is discussed, with stopping criteria such as a minimum change in centroids or reaching a maximum number of iterations. Importantly, each algorithm is rigorously tested with 50,000 function accesses per test, repeated 30 times to account for stochastic variations. Unlike previous works focusing on a single algorithm, this chapter aims to identify the most suitable algorithm for centroid determination in clustering problems. The comparison covers classical and contemporary algorithms, considering their various inspirations, and seeks to understand their effectiveness in solving clustering problems.