OpenAI launches the Decisions API for generating structured, typed outputs from text and image inputs
Feature update Provisional 86% confidence first seen
OpenAI released a new Decisions API in public beta that evaluates text and/or images and returns structured, typed results intended for application workflows such as classification, routing, and triage. The API is positioned as faster than the Responses API and currently uses a specific decisions model accessible via POST /v1/decisions, with pricing set per million input tokens. An accompanying plugin (llm-openai-decisions) was also made available to support image-question queries through the API.