Abstract: 【Purpose/significance】Government open data sharing can accelerate data flow and knowledge diffusion, and deepen the modernization of government governance system and governance capacity.【Method/process】Based on the questionnaire survey, ex⁃ pert scoring method and system architecture design theory, this study identifies the needs, pain points and expectations of 1009 respon⁃ dents for government open data sharing for 31 provinces, autonomous regions and municipalities directly under the central govern⁃ ment, such as Guangdong, Inner Mongolia, Beijing, etc., and constructs a government open data sharing model.【Result/conclusion】 The government open data model based on generative artificial intelligence can integrate data access, integration, screening, download⁃ ing, updating, ease of use, permission and security modules, and promote data standardization and synergy more efficiently and flexibly through the capabilities of natural language processing and multi-round dialogues, solving the realistic problems of data dispersion, high threshold of access, low access efficiency, and so on.【Innovation/limitation】The government open data sharing Q&A service built by integrating the model and OpenAPI specification has the ability of independent management and regular optimization, and is more flexible and controllable than ChatGPT-4.0 in terms of system maintenance, update and cost control. The service can directly in⁃ terface with the original government dataset to provide accurate and real-time government data query, ensuring the maintainability and traceability of the data.