The proliferation of Internet of Things (IoT) networks has increased exposure to sophisticated cyberattacks, demanding robust intrusion detection systems (IDS). This paper presents a unified, cross-model performance analysis of five deep learning (DL) frameworks—Transformer-based IDS, Graph Neural Networks (GNN), Conditional Variational Autoencoder (CTVAE), SimCLR-based contrastive learning, and Federated MLP—under a common preprocessing and evaluation protocol. Experiments on two recent real-world datasets (RT-IoT 2022 and ACI IoT 2023) are used to evaluate the performance matrices. Transformer and SimCLR models achieve up to 99% accuracy on RT-IoT 2022, while Federated MLP excels on ACI IoT 2023, highlighting deployment trade-offs between centralized and privacy-preserving settings. We further discuss computational efficiency, interpretability considerations, and practical deployment guidance, and outline directions for hybrid and lightweight DL on edge devices. Unlike prior works that evaluate single-model IDS frameworks, this study provides the first unified cross-model comparison integrating both centralized and decentralized deep learning paradigms for IoT intrusion detection.