OpenAI Launches Decisions API for Fast Real-Time Classification

At its 2026 Developer Day event in the US, OpenAI announced the Decisions API, built for real-time, low-latency classification and routing. It returns results in about 150 milliseconds, roughly 10 times faster than using the Luna model through a regular API.

The Decisions API is designed for fast single-shot decisions. It focuses the model's intelligence on a set of user-defined questions and limited answer choices, returning structured decisions with very low latency. Developers provide context as text or images, preset questions and possible answers, and the API returns the chosen answer along with confidence scores.

The API fits use cases like customer service ticket classification and priority routing, content moderation policy selection, decision nodes in agent workflows, and real-time business rule assistance. Its main competitor is Jev from TypeSafe.