Decompesition allocation and collaborative optimization of hydrological models are effective technical means for hydrological modeling. Based on literature clustering analysis, combined with bibliometric data and clustering results, the methods of module splitting, combination and optimization of hydrological models were reviewed. The key research progress in aspects such as modular design of hydrological models, framework model construction, and error transmission between multiple models was summarized. The research direction of flexible grouping and collaborative optimization of hydrological models was prospected. It is pointed out that the traditional construction methods of hydrological models usually lack necessary flexibility and portability. The concept of modular design and the integration of cutting-edge technologies such as machine learning have opened up a broad perspective for the future development of hydrological models, which will provide powerful tools to address the complex challenges faced by modern hydrology.