NLP-ADBench presents a comprehensive benchmark for evaluating anomaly detection methods in natural language processing tasks. The benchmark includes diverse datasets, evaluation metrics, and baseline methods, addressing a critical gap in standardized evaluation for NLP anomaly detection research.