OpenAI Makes the Decisions API Generally Available
OpenAI's 22 October note says the Decisions API, on GPT-6 Luna, is generally available at $0.10 per million input tokens.
Direct answer
On 22 October 2026 OpenAI made the Decisions API generally available. It uses GPT-6 Luna, accepts up to 400 questions per request, and costs $0.10 per million input tokens with no output-token charge.
OpenAI's product note dated 22 October 2026 says the Decisions API is generally available. It is built with GPT-6 Luna and is described as focusing that model on questions with bounded answers. The note says this is up to 10 times faster than GPT-6 Luna through the Responses API, for decisions from text, images, and structured inputs.
Since the public beta, the page lists a native GPT-Live integration, prompt caching for reused context, and inference residency in the US and EU. Developers can pass native JSON inputs, include images in question instructions, and ask up to 400 questions per request, which the note calls twice the previous limit. The stated price is $0.10 per million input tokens, and $0.01 per million tokens for cached input. The page says there is no output-token charge. Regional processing is available in the US and EU, and Zero Data Retention is available to eligible customers.
Answer types and the customer figures on the page
The API returns one of three types. A predicate estimates the probability, from 0 to 1, that a condition is true. A choice selects one of the supplied values and also includes probabilities and confidence. A score rates an input against ordered rubric levels and returns the probability-weighted average of the level indices.
The note quotes customer evaluations rather than an OpenAI-wide benchmark. Box says route selection on 30 certificate-of-insurance cases was about 5 times faster than the Responses API, with the same accuracy. Datadog says p95 was 40 times faster than available frontier models, with comparable accuracy, when scoring agent-generated shell commands. Ramp says attribution was 4 times faster and 30 percent cheaper than GPT-6 Luna. Sierra says live orchestration ran under 280 milliseconds at p95, about 3 times faster and 90 percent lower cost per call. Speak says a GPT-Live test cut time to the next action from 1.4 seconds to 0.13 seconds.
FAQ
- Which answer types does the API return?
- The note lists three: a predicate probability from 0 to 1, a choice among supplied values, and a score that is the probability-weighted average of rubric level indices.
- What price does the page state?
- Decisions API costs $0.10 per million input tokens. Cached input costs $0.01 per million tokens. The page says there are no output-token charges.