AI & agents

OpenAI launches the Decisions API for real-time judgment

Published
Source
OpenAI Blog

Summary

OpenAI has made the Decisions API, built on GPT-6 Luna, generally available. It helps applications and agents make small decisions—classifying an input, routing a request, choosing the next action—at near real time, which the company says is up to 10x faster than going through the Responses API. The API supports three answer types (probability, choice, score), up to 400 questions per request, inference residency in the US and EU, and native GPT-Live voice integration.

Why it matters for our work

Agent bottlenecks often come from the many small decisions inside a workflow rather than long reasoning. A cheap, fast path for those decisions can cut the moments where AI keeps people waiting in real-time work such as voice support.

Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.

Read original (opens in a new tab)