Abstract: In recent years, the research on information management based on Grid Information Model (GIM) has increased, and researchers need to effectively visualize and manage GIM model objects at the regional or site level. However, GIM model has the characteristics of large span and high fineness, computers are prone to encounter performance bottlenecks when visualizing and applying GIM models. For this reason, a method of lightweight GIM data is proposed based on BIMBase technology, which first parses GIM file and grids parametric geometry; then removes redundant data in the grid by a merging algorithm with vertex attribute retention; finally, after the GIM model data light weighting is completed, the vertex merging and model loading and parsing performance are compared, and good results are achieved.