OpenAI opens Decisions API public beta for text and image classification
The previously limited API is now available to all developers in public beta, with three typed decision formats and GPT-6 Luna as its only supported model.
A broader release for bounded decisions
OpenAI made its Decisions API available to all developers in public beta on October 6, after previewing it at DevDay in September. The new developer guide documents a dedicated endpoint that uses GPT-6 Luna to evaluate text, images or both. A request can ask for a predicate, such as the estimated probability that an object is damaged; a choice from predefined options; or a score against ordered levels. This is a change in access and documented functionality from the earlier limited preview, not a claim that the API has reached general availability.
Speed claim and operating constraints
OpenAI says decisions can be returned about ten times faster than using GPT-6 Luna through its Responses API. That is a company comparison, not a benchmark independently verified across real applications. The guide lists a base price of $0.10 per million input tokens with no output-token charge, while noting that regional processing and long-context pricing can add costs. GPT-6 Luna is currently the only supported model, and images must be sent as inline base64 data URLs. OpenAI advises developers to test decision thresholds against labeled examples, because a probability or confidence value does not remove the cost of a wrong routing or classification decision.