Accurate cost estimation is a critical factor in the success of construction projects, influencing financial planning, resource allocation, and sustainability. Traditional cost prediction methods often rely on historical data and manual processes, limiting their accuracy and adaptability. To address these challenges, the construction industry is increasingly integrating Building Information Modeling (BIM) and Artificial Intelligence (AI) to enhance cost estimation. This study provides a comprehensive literature review on the state-of-the-art BIM-AI cost prediction models, identifying key methodologies, algorithms, and interoperability challenges. The research systematically analyses 18 highly relevant articles selected from Scopus, focusing on AI-driven cost estimation techniques across various construction lifecycle phases. The review highlights commonly used AI models, including Artificial Neural Networks (ANN), Genetic Algorithms (GA), Deep Learning (DNN, CNN), and Random Forest (RF), evaluating their accuracy and efficiency. The study also explores BIM tools (Revit, Navisworks) and data exchange formats (IFC, gbXML, Excel), identifying interoperability and standardization challenges in AI-BIM integration. Findings reveal that while BIM-AI models significantly improve cost prediction accuracy, their application is largely concentrated in pre-construction and construction phases, with limited research on operation, maintenance, and demolition costs. Future research should focus on developing standardized data exchange protocols, optimizing AI models for cost prediction in later project phases, and integrating real-world validation case studies. This study contributes to advancing digital transformation in construction, promoting AI-BIM methodologies for enhanced financial decision-making and cost management.