Gartner Predicts AI Inference Costs Per Agentic Workflow Will Increase More Than Fivefold Through 2028

As AI products evolve from assistive features to multistep execution, product leaders face a new margin challenge with falling model prices subsidizing more complex workflows and escalating total AI costs. As a result, inference cost management has become a top priority for product leaders.

“Product leaders cannot rely on more efficient token economics to rationalize AI costs,” said Will Sommer, Sr. Director Analyst. “Each successive generation of AI capability will necessitate more, and often more expensive, tokens. There is no reliable, economical one-size-fits-all model on the horizon. Producing competitive AI products will require developing and maintaining complex multimodel ecosystems.”

Gartner has identified three fundamental trends driving token economics:

1. Foundational model cost economics are rapidly improving.
2. Improved AI efficiency is unlocking the deployment of more powerful, and more expensive, models to enable higher-value, more sophisticated AI applications.
3. More sophisticated AI workflows use far more tokens than simple chatbot interactions, driving higher overall inference costs.

These dynamics mean that tokens are becoming more cost-efficient, but not as quickly as AI capabilities and the costs associated with those capabilities are increasing (see Figure 1). The rate of innovation is outpacing the cost curve.

“The harsh economics of the Inference Paradox are exemplified by the differences between a simple chatbot and an AI agent,” said Sommer. “Where a simple chatbot must read and interpret a query and quickly respond with a probabilistically reasonable answer, an AI agent must constantly reason, negotiate, and question itself.”

All of these responsibilities add up. Compared to a basic chatbot interaction, routing a task to an agentic reasoning model increases provider inference costs by at least five times, and often much more as task complexity grows.

Ensuring ROI from advanced AI, like reasoning agents, demands exponentially higher returns relative to basic models, or highly optimized inference-tiering, routing and orchestration to calibrate complex tasks relative to more cost-efficient intelligence. Both of these outcomes are eminently possible but will require significant effort across complex workflows.

“Defaulting to generic autonomous intelligence will result in unbounded costs orders of magnitude higher than those of optimized product ecosystems,” said Sommer.