OpenAI’s Decisions API is GA. Use it for bounded choices, not open-ended work.

OpenAI made the Decisions API generally available on 22 October. A request supplies text, images or native JSON, then asks for a predicate, choice or score. OpenAI says the GPT-6 Luna-based endpoint handles up to 400 questions per request and can run up to ten times faster than Luna through the Responses API. ¹

The listed price is $0.10 per million input tokens and $0.01 per million cached input tokens, with no output-token charge. Regional processing is available in the US and EU; Zero Data Retention is available to eligible customers. ¹

Start with one repeated, bounded decision. Replay labeled production examples for a classifier, router or policy gate; compare accuracy, p95 latency and cost against the current path, and send low-confidence cases to the existing fallback. Keep free-form reasoning and tool execution outside this endpoint.