The expansion in IoT devices has transformed industries, smart cities, and home automation. However, this networked world is very susceptible to increased cyber attacks, thus requiring powerful intrusion detection. Traditional security defenses are not effective since the dynamic nature and high-dimensional data streams in IoT systems are challenging to handle for them. This article proposes a deep learning IDS hybrid, tailored specifically for IoT networks, which uses multiple deep learning models for high detection rates alongside flexibility. It is designed to analyze real-time network packets, identify sophisticated attack signatures, along with providing explainable insights utilizing Explainable AI (XAI). XAI enhances model transparency, allowing security analysts to trace decision paths, thereby promoting trust as well as effective response strategies. In-depth analysis on real-world IoT data demonstrates that the proposed approach performs better than conventional IDS approaches with regard to detection ability, flexibility, and computational efficiency. The current work contributes immensely to IoT security by bridging the gap between performance in deep learning and explainability in IoT model protection against next-generation cyber attacks.