Credit based pricing models are common for AI and agenticAI companies. Design of these models is becoming a critical skill for go-to-market and pricing teams. One of the key question when designing this model is how granular to make the credits. There is a tension between tracking cost and value and between simplicity and precision. Managing these tradeoffs is critical to the design of credit based pricing models.

Tokens and credits are becoming a popular way to price AI agents and applications. Why is this? They offer a unique combination of predictability that is important when pricing early stage general purpose technologies or when there are too many use cases for conventional value based pricing. Design of credit or token based pricing requires special attention to balancing the different pricing incentives.