Dynamic pricing mechanisms for generative artificial intelligence models across heterogeneous scenarios: An evolutionary game and complex network approach

Generative artificial intelligence (GAI) models have shown significant versatility and reusability in applications such as natural language generation, image synthesis, and code completion. However, when applied across industries, these models face challenges due to differences in model performance, risk tolerance, and price sensitivity. The "one-size-fits-all" pricing model not only increases the administrative burden on model providers but also diminishes the willingness of users to adopt these technologies. This paper views GAI models as a new type of data asset, combining traditional data characteristics with unique attributes. To address this, we develop a dynamic pricing framework that combines evolutionary game theory and complex network theory. By incorporating a Q-learning algorithm, a model for edge disconnection and reconnection, and a trust-based dynamic mechanism, we model the strategy evolution of both providers and demanders in the face of information asymmetry, bounded rationality, and heterogeneous network connections. The simulation results indicate that cooperation rates tend to follow an "inverted U-shaped" evolution. Among the factors influencing long-term cooperation, the continuous usage fee is found to be the most sensitive parameter. Additionally, the study examines how factors like customization costs, company size, and scenario-specific needs shape pricing strategies and market cooperation. The results highlight the importance of differentiated and flexible pricing strategies for fostering sustainable cooperation. These findings provide both a theoretical framework and empirical evidence to guide the development of adaptive and sustainable pricing strategies for GAI models.