The prediction of diabetes is a challenging task due to the complex and multifactorial nature of the disease. In recent years, machinelearning algorithms have been applied to predict the onset of diabetes using various sets of predictors, such as demographic, clinical, and laboratory data. In this study, we propose a firefly algorithm to identify diabetes and compare its performance with other algorithms. We evaluate the performance of the firefly algorithm using four widemetrics for evaluation:accuracy, precision, recall, and F-score. Our experiments were conducted on a real-world dataset consisting of 768 individuals, of which 268 had diabetes. The training and testing sets were randomly divided into two groups with an 80:20 ratio. We performed the firefly algorithm for feature selection. It is one of the Nature-Inspired Algorithms (NIA). It is used to optimize the parameters using the firefly algorithm. Then the optimized parameters were then used to train the firefly algorithm on the entire training set.The experimental results demonstrate that the firefly algorithm achieves competitive performance compared to other machine learning algorithms in terms of precision, accuracy, F-score, and recall, the firefly method outperforms other algorithms.