OpenAI · 22 October 2026

A faster lane for the small decisions inside an app.

Summary of OpenAI’s own post. The claims below are theirs.

OpenAI says the Decisions API is generally available. It is built on GPT-6 Luna, and it is for questions that already have a bounded answer: classify an input, route a request, or pick the next action. They say those answers come back up to ten times faster than asking the same model through the Responses API.

Since the public beta, they say they have added a GPT-Live integration, caching when the same context is reused, and the option to keep inference in the US or the EU. A request can include JSON and images, and up to 400 questions, which they say is twice the previous limit. They also say they added capacity in several parts of Asia.

This is not a chat. It is a pile of small, already-shaped choices.

If the app in your head is full of those choices, the plan still starts in the chat you already use. The IDE is the part on your Mac: you see the blueprint, you accept it or you don’t, and you watch a preview.