OpenAI Adopts Outcome-Based Pricing for AI Services
OpenAI has started letting some customers pay only when its AI successfully completes a task, a strategy known as outcome-based pricing. This shift comes as a significant development in the AI industry, with OpenAI, one of the largest model vendors, embracing a model previously adopted by customer support services.
A Growing Trend in Customer Support
Outcome pricing has been gaining traction in customer support for some time now. Companies like Intercom and Zendesk have already implemented this approach. Intercom, for instance, charges $0.99 for each conversation its Fin agent successfully resolves. Zendesk, in May, introduced a system where billing is restricted to what they call Verified Resolutions, with a rate of roughly $1.20 to $1.50 on committed volume.
The Shift in Billing Models
Salesforce, too, has been exploring this concept, launching Agentforce at $2 per conversation, and later introducing Flex Credits, which charges based on individual actions, starting at $500 for 100,000 credits. While these models represent a move away from traditional per-user pricing, they still bill for actions, even if they fail.
Market Preferences and Risks
According to Futurum Group’s research, a significant majority of buyers prefer consumption-based models (43%) over outcome-based ones (27%), but both are gaining popularity. OpenAI’s decision to adopt outcome pricing is a strategic move, reflecting the company’s confidence in its AI’s success rate. However, it also shifts the risk of poor answers from customers to the AI provider, which needs to be factored into the pricing.
In summary, OpenAI’s new pricing model represents a significant change in the AI industry, with potential implications for both customers and providers.