OpenAI makes the Decisions API generally available
GPT-6 Luna answers bounded questions through a dedicated endpoint. OpenAI prices input at $0.10 per million tokens, cached input at $0.01, and charges nothing for output tokens.
Official source (opens in a new tab)
OpenAI's 22 October 2026 post makes the Decisions API generally available. It uses GPT-6 Luna on a dedicated Decisions endpoint for questions with a bounded answer, from text, images, or native JSON. OpenAI says it is up to 10 times faster than GPT-6 Luna through the Responses API. A request can include up to 400 questions, twice the public-beta limit. Since the beta, OpenAI has added a native GPT-Live integration, prompt caching for reused context, and inference residency in the US and EU, plus more capacity in several regions in Asia.
The API returns one of three answer types. A predicate is a probability from 0 to 1 that a condition is true. A choice is one of the values you supplied, with probabilities and a confidence. A score is the probability-weighted average of ordered rubric levels. OpenAI says you pay only for input tokens: $0.10 per million, or $0.01 per million for cached input. Cached writes at 30 minutes and 12 hours are on the pricing guide, which this post points to rather than reprinting. Zero Data Retention is available to eligible customers.
Who it's for: an application that classifies, routes, or scores the same kind of decision many times and needs that step to stay cheap.
Fits: Best for Writing stays Claude for a draft a person reads. This is an API for a bounded choice, not the chat product. The verdicts are unchanged.