This study evaluates the robustness and performance of graph-based models for credit default prediction. Two inductive graph neural network (GNN) architectures – GraphSAGE and the graph attention network (GAT) – are implemented within a framework that integrates automated hyperparameter optimization, imbalance-aware loss functions and adversarial stress testing. Borrowers are represented as nodes in a k-nearest neighbor graph constructed from financial and demographic features. Model tuning is performed via Optuna, while robustness is examined under the fast gradient sign method and projected gradient descent perturbations, with adversarial training enhancing stability. Experimental results demonstrate that optimized and adversarially trained GNNs outperform classical baselines such as logistic regression, random forest and gradient boosting in area under the curve and F₁ metrics, while maintaining resilience under feature perturbations. These findings highlight the importance of robustness evaluation as part of the broader model assessment process for modern credit risk modeling.