<p>Since 2016, the Chinese government has issued a series of policies, laws and regulations, providing a top-level design for big data governance. This “big data-based governance” model represents a significant theoretical innovation, as well as a notable advancement in both policy and practice within the domain of public administration. The model emphasises the utilisation of big data-related information technology by government agencies for the cataloguing of government data production, management and collection, as well as the flow of data across governmental departments. Additionally, it seeks to enhance the government’s social governance through the use of this technology. This paradigm shift in government operations, underpinned by digital transformation, signifies a paradigm shift in the realm of public administration and services. The integration of big data as a critical component of governance facilitates the enhancement of government functions across all levels of governance. The government is able to upgrade, innovate and respond to demands by using big data as a means of governance. The present paper puts forward the argument that such a practice presents a theoretical model of big data-based governance, where the focus is placed on the production, management and collection of government data, the flow of data across governmental departments and social areas, and the technical upgrading of governmentality through the use of information technology tools related to big data. To elucidate this definition, the paper presents a topological analysis of the capabilities of this “big data-based governance” model, proposing four orders of capabilities to encapsulate the essence of this model. The “zero-order capability” is primarily reflected in the establishment of the national big data platform for government affairs; the “first-order capability” refers to the various types of data resources for government affairs; the “second-order capability” is the development of various data-based application services and systems on the basis of the data resources of the “first-order capability”; finally, the “third-order capability” is future-oriented, which can be termed as the “fifth paradigm of governance”. The concept encompasses the utilisation of big data, algorithms and artificial intelligence to train a large governance model (LGM) based on big data. This model can facilitate the implementation of certain mechanisms that enable autonomous decision-making.</p>