AI Catchup

OpenAI's Decisions API Turns Text and Images into Fast, Typed Answers

By 3 min read

OpenAI's Decisions API is a public-beta endpoint for asking structured yes/no, classification, and scoring questions about text or images. It currently uses GPT-6 Luna, and OpenAI says it returns typed answers about 10x faster than the Responses API. The endpoint charges $0.10 per million input tokens, with no cache-read, cache-write, or output-token fees.

OpenAI released the Decisions API in public beta on October 6, 2026. The dedicated POST /v1/decisions endpoint currently uses gpt-6-luna to evaluate text, images, or both. It returns typed answers for condition checks, fixed-choice selection, and rubric scoring. (OpenAI API changelog; Decisions API guide)

OpenAI says the endpoint returns typed answers about 10x faster than the Responses API. It is built for applications that need a decision, such as whether a condition is true, which option fits, or how an input scores against a rubric, rather than a free-form explanation. (Decisions API guide)

Three kinds of decisions

Each request provides shared evidence and one or more questions about it. Independent questions can go in the same questions array; when one decision depends on an earlier answer, OpenAI's guide says to send a separate request. (Decisions API guide)

Question typeWhat it returnsExample use
predicateA probability from 0 to 1 that a condition is trueCheck a product image for visible damage
choiceOne of the supplied optionsRoute a customer complaint to a department
scoreA score calculated against ordered levelsRate issue severity using a defined rubric

For a choice question, describe distinct options and include a fallback if the categories may not cover every input. For a score, define ordered levels with clear criteria. OpenAI recommends writing questions around observable criteria and keeping separate concerns in separate questions. (Decisions API guide)

Where it fits in an application

The endpoint is aimed at repeated decision tasks such as classification, routing, prioritization, and rubric-based evaluation. It lets an application ask a model for a typed answer without treating every request as a general response-generation task. (Decisions API guide)

It is not the right fit for every structured response. OpenAI recommends Structured Outputs with the Responses API when an application needs an object that follows its own JSON schema, such as extracted fields or a written explanation. For a model-generated tool request, OpenAI points developers to function calling instead. (Decisions API guide)

The endpoint accepts text and image inputs. Check OpenAI's current API reference for the request schema and supported input forms before deployment. (Decisions API guide; Create a decision API reference)

Beta access and pricing

The Decisions API is in public beta, and OpenAI says it expects general availability in the coming weeks. The guide lists gpt-6-luna as the only model currently available through the endpoint. (Decisions API guide)

For requests to /v1/decisions, GPT-6 Luna input costs $0.10 per million tokens. OpenAI says the endpoint charges only for input tokens, with no cache-read, cache-write, or output-token charges. Regional-processing premiums and long-context input multipliers still apply; the guide says other requests using GPT-6 Luna follow the pricing for their applicable endpoint and processing tier. (Decisions API guide)

Bottom line

The Decisions API gives developers a focused interface for fast classification and scoring over shared text or image evidence. Test it when an application needs probabilities, a choice from known options, or a rubric score; keep the Responses API for generated explanations or custom-schema outputs. OpenAI's published latency and pricing details are specific to this endpoint and its current beta scope. (Decisions API guide)

Sources

Frequently Asked Questions

What is OpenAI's Decisions API?

The Decisions API is a public-beta endpoint for asking classification and scoring questions about text, images, or both. It returns typed answers such as a probability, a choice from fixed options, or a score against ordered levels.

Which model does the Decisions API use?

OpenAI's guide says GPT-6 Luna is the only model currently available through the Decisions API.

How much does the Decisions API cost?

For the Decisions API, GPT-6 Luna input costs $0.10 per million tokens. OpenAI says it charges only for input tokens, with no cache-read, cache-write, or output-token charges. Regional processing premiums and long-context input multipliers still apply.

Get the weekly AI Catchup

Tools, practices, and what matters, in your inbox every week.