OpenAI Just Dropped a New Decisions API That Forces AI to Finally Make Up Its Mind
Because vague philosophical poetry from chatbots isn't corporate enough, OpenAI launched a dedicated endpoint for hard math, probabilities, and definitive choices.
During their recent developer event, OpenAI rolled out a specialized tool called Decisions API, running exclusively on the gpt-6-luna model. The service ditches free-form text generation in favor of rigid, structured outputs, calculating exact probabilities, category selections, and numerical ratings for both text and images.
Instead of asking a model to write a paragraph about a photo, developers can now submit targeted questions broken down into three specific types. A predicate checks the mathematical likelihood of a condition being true, a choice picks a category from a predefined list, and a score ranks inputs along an ordered numerical scale. Multiple independent questions can be fired at the same dataset in a single request, returning confidence metrics alongside the raw percentages.
Pricing for the endpoint is set at ten cents per million input tokens, with zero fees charged for output tokens or cache operations. This positions the tool in direct competition with Jev by TypeSafe AI, which offers a lower base price but lacks native image processing capabilities out of the box.
Probabilistic certainty has officially become a utility bill, transforming human judgment into a background API call that costs fractions of a cent while outsourcing corporate responsibility to a statistical cluster.
Source: OpenAI DevDay
Comments
Help shape the next version: Add context or suggest a correction. AI review can add points toward a rewrite. Reviews and updates may take time; a full meter does not guarantee a new version.