OpenAI Decisions API is generally available with GPT-6 Luna
OpenAI announced that the Decisions API is generally available as of October 22, 2026. The API is built with GPT-6 Luna and is designed to make bounded, near-real-time decisions from text, images, and structured inputs, reportedly up to 10x faster than GPT-6 Luna through the Responses API. The release matters because applications and agents make many small decisions throughout a workflow, and their combined latency and cost add up at scale.
What the Decisions API does
OpenAI describes the Decisions API as a way to focus GPT-6 Luna's intelligence on questions with bounded answers. Developers define a question and its allowed answers, then provide context as text, images, or native JSON. The API returns one of three supported answer types: a probability, a choice, or a score. These answer types are intended to cover a wide range of use cases where an application needs a fast, structured decision.
| Type | Use it to | Main result |
|---|---|---|
| predicate | Estimate the probability that a condition is true, such as whether a product has visible damage. | probability: an estimate from 0 to 1 that the condition is true. |
| choice | Select one of your predefined options, such as whether an agent should continue, retry, or ask for clarification. | choice: one of your supplied values. The response also includes probabilities and confidence. |
| score | Rate an input against ordered levels in a rubric, such as the severity of damage. | score: the probability-weighted average of the level indices. |
Changes since the public beta
Since the public beta, OpenAI added a native GPT-Live integration, prompt caching for reused context, and inference residency in the US and EU. Developers can also pass native JSON inputs, include images in question instructions, and ask up to 400 questions per request, which is twice the previous limit. OpenAI says it has continued to improve latency, including additional capacity in several regions in Asia.
Voice and live conversation use cases
The Decisions API integrates natively with GPT-Live, OpenAI's voice offering for natural, expressive conversation with smooth interruption handling. The integration is intended to help applications make fast, focused decisions during live conversations. For example, a customer support agent can decide whether to ask a clarifying question, route a request, or escalate an issue. An avatar can adjust its facial expressions as a user speaks and, based on the user's words, respond with a wink or a smile.
The example in the announcement uses transcript updates to select an avatar's expression. Developers add configuration to session.triggers in an existing GPT-Live session. Live emits a decision.event; the application reads the selected reaction and animates the character accordingly.
How teams are using the Decisions API
OpenAI says early adoption has shown applicability across a variety of use cases. Across OpenAI, teams are applying the Decisions API to product and business workflows. The dots team is using it to reduce response times and improve interactivity. The Personal Finance team is using it to categorize transaction descriptions. The Communications team is using it to classify coverage by themes and scores whether it reinforces or challenges key narratives, helping identify what warrants closer review.
Customers are applying the Decisions API broadly as well, including accurately attributing AI costs and making language-learning experiences more interactive.
Availability, pricing, and data controls
The Decisions API is generally available with GPT-6 Luna through the dedicated Decisions endpoint. Pricing is $0.10 per million input tokens. Cached input costs $0.01 per million tokens. For 30-minute and 12-hour cached writes, OpenAI directs developers to its pricing guide. Only input tokens are billed; there are no output-token charges.
Regional processing is available in the US and EU. Zero Data Retention is available to eligible customers. OpenAI points to its pricing and caching guide and data controls documentation for details.
Getting started
OpenAI recommends starting with one recurring decision an application makes often. Developers should define the question, the possible answers, and the context needed to distinguish them. They can test the decision on examples that reflect real usage in the Decisions Playground with their own text or images, or follow the developer guide to make a first request.