CityCraft is an innovative framework designed to enhance the diversity and quality of 3D city scene generation. It integrates three key stages: layout generation using a diffusion transformer (DtT) model, strategic urban planning with a Large Language Model (LLM), and high-quality asset integration and scene construction using Blender. The framework addresses the limitations of existing methods, which often produce monotonous and unrealistic scenes due to a lack of planning capabilities and diversity. CityCraft introduces two new datasets: CityCraft-OSM, which includes 2D semantic layouts, satellite images, and detailed annotations, and CityCraft-Buildings, which features diverse 3D building assets. Evaluations show that CityCraft outperforms state-of-the-art methods in terms of realism, diversity, and user preference, making it a valuable tool for urban planning, architecture, and virtual reality applications.CityCraft is an innovative framework designed to enhance the diversity and quality of 3D city scene generation. It integrates three key stages: layout generation using a diffusion transformer (DtT) model, strategic urban planning with a Large Language Model (LLM), and high-quality asset integration and scene construction using Blender. The framework addresses the limitations of existing methods, which often produce monotonous and unrealistic scenes due to a lack of planning capabilities and diversity. CityCraft introduces two new datasets: CityCraft-OSM, which includes 2D semantic layouts, satellite images, and detailed annotations, and CityCraft-Buildings, which features diverse 3D building assets. Evaluations show that CityCraft outperforms state-of-the-art methods in terms of realism, diversity, and user preference, making it a valuable tool for urban planning, architecture, and virtual reality applications.