OpenAI has begun letting some of its largest customers pay only when its AI actually completes the job. This arrangement, known as outcome-based pricing, is limited to select major accounts and has not been publicly announced. The terms, customers, and prices are all unknown, but the move is seen as a significant shift in how OpenAI sells its services.
The industry name for this pricing model is outcome-based pricing, and its appeal to finance directors is clear. A bill that arrives only when something worked is easier to defend than one that arrives regardless. Token billing has made this a live problem, with one developer accumulating $1.3m in OpenAI tokens across thirty days, highlighting the need for a more results-driven approach.
Customer support is one area where outcome-based pricing has already taken hold. Intercom charges $0.99 for each conversation its Fin agent resolves, and nothing for the ones it does not. Zendesk has also adopted a similar approach, restricting billing to what it calls Verified Resolutions, confirmed by an LLM evaluation within 72 hours of the conversation.
Salesforce has been working through the same question, initially launching its Agentforce at $2 per conversation, charged for every 24-hour session whether or not anything was resolved. However, customers found this expensive and impossible to forecast, leading to the introduction of Flex Credits, which move the meter from conversations to individual actions at about 10 cents each.
Buyers appear to want both consumption-based and outcome-based models, with 43% preferring consumption-based models and 27% preferring outcome-based ones, according to a survey by Futurum Group. The firm's research director, Keith Kirkpatrick, wrote that outcome-based pricing is becoming a market standard, and vendors offering seats alone are now being disqualified before the evaluation starts.
For OpenAI, the move is a change of position rather than a new product. It has sold capacity by the token, priced per model and per call, and letting an enterprise pay for completed work instead means accepting the risk that the work does not complete. This risk has to be priced somewhere, and the interesting question is where. A vendor confident in its success rate can afford the arrangement, while one that is not has to load the per-success price until the economics match.
The harder part is agreeing what success means. A resolution is definable, but the agentic work OpenAI has been pushing towards involves multi-step tasks where completion is a matter of judgement rather than a field in a database. There is a commercial reason to want it settled quickly, as an enterprise that cannot forecast a bill tends to run a pilot indefinitely rather than sign, and outcome pricing removes the objection at exactly the point in the sales cycle where it usually stalls.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.