# OpenAI Makes Decisions API Generally Available to Developers By Simon Yoon Canonical URL: https://www.tokenpost.com/news/technology/38976 Published: 2026-10-23T12:50:20.000Z Updated: 2026-10-23T12:50:20.000Z Section: Technology > Requests can include up to 400 questions, double the beta limit. The release adds image inputs, prompt caching and inference residency in the U.S. and EU. OpenAI made its Decisions API generally available Oct. 22, giving software developers a GPT-6 Luna tool for tasks that need fixed outputs rather than open-ended text. For bounded questions, the endpoint processes text, images or structured inputs and returns a probability, selected option or score. Developers can use those responses in workflows that sort support requests, route documents or decide whether an item should be escalated to a person. The format is designed for narrow questions where software needs an answer it can process directly. A request can now contain up to 400 questions, double the 200-question limit during beta. OpenAI opened the [public beta on Oct. 6](). The API currently runs on GPT-6 Luna, and the general-availability release adds native GPT-Live integration, prompt caching and inference residency in the U.S. and EU. OpenAI lists base input prices of $0.10 per million uncached tokens and $0.01 per million cached tokens, with no charge for output tokens. Hsu said Speak’s GPT-Live testing cut the time before the next action from 1.4 seconds to 0.13 seconds. He described the result as coming from Speak’s testing of its integration, rather than a general benchmark across workloads. The API’s general availability gives developers broader access to a tool for constrained decisions, with image and structured-data support alongside text. The release also expands the beta’s request limit and adds options for caching and regional inference. ## Links in this article - [public beta on Oct. 6](https://www.tokenpost.com/news/technology/28430)