Nonlinear trade-offs in GenAI application pricing: a game-theoretic analysis of API and psychological costs

In the rapidly evolving generative artificial intelligence (GenAI) ecosystem, downstream application providers face persistent challenges in designing optimal pricing strategies. These challenges arise from dual cost pressures: application programming interface (API) costs imposed by upstream model platforms and psychological frictions experienced by end consumers. This study develops a game-theoretic model to analyze application providers’ strategic choice between subscription-based and usage-based pricing under both monopolistic and duopolistic market structures. The analysis reveals that the optimal strategies are shaped by a nonlinear interplay among psychological costs, model API costs, and consumer valuation. Under low psychological-cost conditions, API costs exert an inverted U-shaped effect on competing providers’ profits. By contrast, under high API-cost regimes, subscription-based and usage-based models exhibit asymmetric sensitivities to psychological costs: profits under usage-based pricing decline with rising psychological costs, whereas subscription-based profits stabilise. Fthermore, elevated API and psychological costs attenuate competitive intensity, driving duopolistic equilibria toward monopoly-like outcomes. These findings remain robust after incorporating a broad set of realistic market frictions. The results provide actionable insights for GenAI application providers balancing cost pass-through, user engagement, and competitive positioning in pricing decisions.