This paper highlights the importance of the Foundation model in the IoT domain, which reduces labeled data dependency and demonstrates strong generalization across tasks. We note that existing Foundation model-based methodologies are limited to specific IoT tasks, making it difficult to compare them across IoT domains and provide guidelines for applying them to new tasks. To address this gap, we provide a comprehensive overview of current methodologies. The paper organizes the methodologies around four shared performance goals—efficiency, situational awareness, safety, security, and privacy—and reviews representative research, commonly used technologies, and evaluation metrics for each goal. This goal-oriented organization enables meaningful cross-domain comparisons and provides practical insights for selecting and designing Foundation model-based solutions for new IoT tasks. Furthermore, we suggest future research directions for advancing the use of Foundation models in IoT applications.
