Abstract: GRSPES (Global and Regional Assimilation and PrEdiction System) is a new-generation general hydrostatic/non-hydrostatic, multi-scale numerical model of China developed by Research Center for Numerical Meteorological Prediction of CAMS (Chinese Academy of Meteorological Sciences) of CMA. The Model adopts a structure of standardized and module-based software and is developed in accordance with the strict requirement of software engineering, including the program-operated parallel calculation. Preliminary calculations show that the design and application of the model agree on the pre-requirements, thereby laying a good foundation for the sustainable development of the numerical prediction system of the country. Several approaches in GRAPES model: standardization, module and multi-scale community. According to the principles of software engineering, the system and its parallel computing are developed. Results of experiments show that both the design of model framework and the implementation of GRAPES model can meet the requirements of model processing. It's asserted that this framework plays an important role in GRSPES development.