<p>This paper examines the application of hybrid metaheuristic optimization algorithms for clustering Arabic data, aiming to distinguish between similar and dissimilar groups effectively. Metaheuristic optimization, particularly swarm intelligence, has recently provided solutions to the limitations of traditional clustering algorithms such as K-means, which, while practical, often suffer from issues such as becoming trapped in local optima and having slow convergence rates, especially when handling high-dimensional data. Various successful metaheuristic optimization algorithms and swarm intelligence-based methods have been developed to address these limitations and find optimal solutions within a reasonable timeframe. This study introduces a robust approach using hybrid optimization algorithms, gray wolf optimization and cuckoo search optimization, with the minibatch K-means method. We evaluated the overall performance of the proposed method in four Arabic document datasets—ARPD, NADA, BBC, and CNN. The results are compared with those of traditional clustering methods, other metaheuristic optimization algorithms, and recent cutting-edge clustering techniques. The experimental findings demonstrate that our proposed method significantly outperforms existing techniques by maximizing the silhouette coefficient and the adjusted Rand index, achieving the lowest root-mean-square error and acceptable computational speed.</p>