The paper introduces the Graph Kolmogorov–Arnold Network (GKAN), a novel Graph Neural Network (GNN) model that leverages spline-based activation functions on edges to enhance both accuracy and interpretability. GKAN is designed to address the lack of interpretability in GNNs, which is a significant issue in domains requiring transparent decision-making. The model's architecture includes KAN-based convolutional and linear layers, allowing for flexible nonlinear transformations of node features based on their connections. Experiments on five benchmark datasets (Cora, PubMed, CiteSeer, MUTAG, and PROTEINS) demonstrate that GKAN outperforms state-of-the-art GNN models in node classification, link prediction, and graph classification tasks. Additionally, GKAN provides inherent interpretability by design, eliminating the need for external explainability techniques. The paper discusses the methodology, performance, and interpretability of GKAN, highlighting its potential for applications in fields where interpretability is crucial. Future work includes optimizing computational efficiency, incorporating edge features, and exploring real-world applications.The paper introduces the Graph Kolmogorov–Arnold Network (GKAN), a novel Graph Neural Network (GNN) model that leverages spline-based activation functions on edges to enhance both accuracy and interpretability. GKAN is designed to address the lack of interpretability in GNNs, which is a significant issue in domains requiring transparent decision-making. The model's architecture includes KAN-based convolutional and linear layers, allowing for flexible nonlinear transformations of node features based on their connections. Experiments on five benchmark datasets (Cora, PubMed, CiteSeer, MUTAG, and PROTEINS) demonstrate that GKAN outperforms state-of-the-art GNN models in node classification, link prediction, and graph classification tasks. Additionally, GKAN provides inherent interpretability by design, eliminating the need for external explainability techniques. The paper discusses the methodology, performance, and interpretability of GKAN, highlighting its potential for applications in fields where interpretability is crucial. Future work includes optimizing computational efficiency, incorporating edge features, and exploring real-world applications.