OpenAI launches Decisions API for fast in-app routing and tool selection

OpenAI launches Decisions API for fast in-app routing and tool selection

N
News Editor
2026-09-30 01:06:18
OpenAI introduced the Decisions API at DevDay as a product built for quick, bounded choices inside software rather than long-form text generation. Developers provide a fixed set of options, then pass in text or images, and the underlying GPT-6 Luna model returns a decision in a few hundred milliseconds. The reported use cases include routing a message to sales or customer support and letting an agent choose which tool to call next. According to BlockBeats, similar tasks could already be handled by general-purpose large language models, but that usually meant generating text first and then parsing the output in code. The new API instead constrains Luna to a preset answer set to speed up frequent micro-decisions. The report also noted similarities to the recently popular Jev, though it drew a distinction between the two: Jev is described as a purpose-trained decision model, while OpenAI is currently offering constrained decision-making on top of Luna, with added image input support. Decisions API is in limited preview for now, and OpenAI plans to expand access in the next few days.

OpenAI unveiled its Decisions API at DevDay, pitching it as a tool for fast choices inside software rather than long-form response generation.

Developers can define a small set of explicit options, then send in text or images. The underlying GPT-6 Luna model makes a decision in a few hundred milliseconds.

Built for repeated, narrow decisions

BlockBeats said the API is aimed at common decision points in applications, such as deciding whether a message should go to sales or customer support, or choosing which tool an agent should call next.

Those tasks could previously be handled by a general large model, but that often required generating a text response first and then parsing the result in code. With Decisions API, Luna is constrained to a fixed list of candidate answers, with the goal of making these frequent small judgments faster.

Comparison with Jev

The design invites comparisons with Jev, which has drawn attention in recent weeks. The report said TypeSafe also shifts AI from text generation to direct choice-making.

Still, the two are not identical. Jev is described as a purpose-trained decision model, while OpenAI is currently using Luna for constrained decisions and also supports image input.

Limited preview for now

Decisions API remains in limited preview, and OpenAI plans to broaden availability over the next few days.

This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan.
1500

Disclaimer:

The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.

Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.