OpenAI made its Decisions API generally available on Oct. 22, 2026, after opening a public beta to all developers on Oct. 6. The API is designed for small decisions inside apps and agents, such as classifying an input, routing a request or choosing what to do next.
A question with limited answers
Developers define a question and its allowed answers, then supply context as text, images or native JSON, according to OpenAI. The API returns a probability that a condition is true, a choice from supplied options or a score against a rubric. The app can use that result to decide its next step. OpenAI describes the approach as making a decision without generating a full model response.
OpenAI says the GPT-6 Luna-based API handles these bounded questions up to 10 times faster than GPT-6 Luna through the Responses API. That is the company’s comparison with a specific way of using Luna, not a speed figure for every app.
What changed since beta
OpenAI lists native GPT-Live integration for voice apps, prompt caching for reused context, and inference residency in the US and EU among the additions since beta. Developers can also provide native JSON inputs, put images in individual question instructions and ask up to 400 questions per request, twice the previous limit. OpenAI says it has added request capacity in several regions in Asia.
For a concrete voice use case, OpenAI’s article quotes Speak CTO Andrew Hsu as saying that testing the Decisions API with GPT-Live reduced the time to next action from 1.4 seconds to 0.13 seconds for its language-learning app. That figure describes Speak’s test, while OpenAI’s broader speed comparison uses GPT-6 Luna through the Responses API.
OpenAI’s Oct. 22 article listed a price of $0.10 per million input tokens, $0.01 per million cached input tokens and no output-token charges. It directs developers to its pricing guide for current rates and caching details.
Comments