• Introduces MltAuxTSPP, a unified and scalable benchmark framework for deep learning-based traffic state prediction with multi-source auxiliary data.
• Features a standardized data container and fusion embedding module, enabling seamless integration of heterogeneous data sources.
• Facilitates fair and reproducible comparisons of downstream models under identical conditions, addressing the lack of unified evaluation.
• Demonstrates that leveraging weather and temporal features improves long-term forecast accuracy, offering a practical foundation for ITS research.