#

Key takeaways

• OpenAI says the Decisions API is now generally available.

• It is built with GPT-6 Luna and targets bounded-answer decisions.

• OpenAI says it can be up to 10x faster than GPT-6 Luna through the Responses API.

• The update adds GPT-Live integration, prompt caching, and native JSON inputs.

• Developers can ask up to 400 questions per request.

What OpenAI announced

OpenAI announced that the Decisions API is generally available. The company says the API is designed for fast, focused, low-cost decisions inside applications. It is meant for tasks where applications and agents make many small choices during a workflow.

Those choices can include classifying an input, routing a request, or selecting the next action. OpenAI says each decision is simple on its own. At scale, however, latency and cost can add up. The Decisions API is positioned as a way to handle those decisions more efficiently.

How the API works

OpenAI says the Decisions API is built with GPT-6 Luna. It focuses Luna's intelligence on questions with bounded answers. The goal is to help applications and agents make near real-time decisions from text, images, and structured inputs.

OpenAI says the API can be up to 10x faster than GPT-6 Luna through the Responses API. That comparison is part of the product framing in the source material. The company also says developers can use native JSON inputs and include images in question instructions.

The request limits also changed. Developers can ask up to 400 questions per request. OpenAI says that is twice the previous limit. The source also says the API now includes prompt caching for reused context.

What changed since public beta

OpenAI says several features were added since public beta. These include a native GPT-Live integration, prompt caching, and inference residency in the US and EU. The company also says it continued to improve latency.

OpenAI mentions additional capacity in several regions in Asia. The source does not provide more detail on those regions. It only states that capacity was added and latency improved.

How teams are using it

OpenAI says early adoption has shown the API can support a variety of use cases. The source says teams across OpenAI are applying it in product and business workflows. It gives two examples.

The dots team is using the Decisions API to reduce response times and improve interactivity. OpenAI's Personal Finance team is using it to categorize transaction data. These examples show the API being used for workflow decisions rather than open-ended generation.

Operational considerations

The source points to a few practical considerations. First, the API is designed for bounded answers, so it fits decision tasks better than broad, open-ended ones. Second, the value proposition depends on latency and cost, since OpenAI highlights both as workflow concerns.

The source also suggests that input format matters. Developers can use text, images, and structured inputs. They can also pass native JSON inputs and reuse context through prompt caching. Those features may matter when teams want consistent decision flows.

Morocco relevance

The source reports no Morocco-specific fact. The global lesson is conditional: if a team is building applications with many small workflow decisions, a bounded-answer API may help reduce latency and cost.

Bottom line

OpenAI is making the Decisions API generally available with a stronger feature set than during public beta. The update centers on speed, bounded-answer decision-making, and broader input support. The source frames it as a tool for applications and agents that need many fast decisions inside a workflow.