█

AI/TLDR

OpenAI · 2026-09-29 · major

OpenAI Decisions API — GPT-6 Luna picks from your answers in 150 ms

OpenAI's Decisions API, launched at DevDay 2026, runs a special GPT-6 Luna that picks one of a developer's predefined answers in about 150 ms, versus 1.6 seconds for a normal Luna call. It is in limited preview.

FourWeekMBA graphic explaining OpenAI's Decisions API and its fixed output space
FourWeekMBA

OpenAI's answer to Jev: an endpoint that returns a choice, not text, about ten times faster than a normal Luna call.

Quick facts

MakerOpenAI
ModelSpecialized GPT-6 Luna
LatencyAbout 150 ms (vs 1.6 s for a standard Luna call)
InputText or image context + a closed list of answers
OutputOne chosen answer with a confidence score
AvailabilityLimited preview; broad rollout in the coming days

What is it?

The Decisions API is a new OpenAI endpoint, announced at DevDay on September 29, 2026, that does not write free text. The developer sets a question and a fixed list of possible answers, and the API returns which answer fits best. The New Stack and The Decoder frame it as OpenAI's response to Jev, the typed decision model TypeSafe AI released in mid-September.

How does it work?

A specialized version of GPT-6 Luna reads the context — text or an image — and scores only the allowed options, then returns the most likely one with a confidence score. Because the output space is closed, the model cannot answer outside the list. Skipping text generation cuts the response time to about 150 milliseconds, against 1.6 seconds for a standard GPT-6 Luna call.

Why does it matter?

Classification, routing and "what should the agent do next" steps are often the slowest and least reliable part of an LLM app, because the model writes text that then has to be parsed. A 150 ms endpoint that can only return a valid option makes these steps fast enough for real-time use. OpenAI has not published a price or an accuracy figure yet, so teams should benchmark it against Jev-style models before switching.

Who is it for?

developers building classifiers, routers and agent control loops

Frequently asked questions

How much does the OpenAI Decisions API cost?
OpenAI has not published a price for the Decisions API. FourWeekMBA notes that OpenAI also gives no accuracy figure yet. For reference, it cites third-party OpenRouter pricing for GPT-6 Luna of $0.10 per million input tokens and $0.50 per million output tokens, but that is Luna's general rate, not a Decisions API price.
Can I use the OpenAI Decisions API today?
The OpenAI Decisions API launched on September 29, 2026 in limited preview, according to The Decoder and FourWeekMBA. OpenAI says a broad rollout will follow in the coming days. It is not generally available yet, so most developers will need to wait for the wider release or request preview access.
How does the OpenAI Decisions API compare to Jev?
Both the OpenAI Decisions API and TypeSafe AI's Jev answer a question by choosing from predefined options instead of generating text. The New Stack describes the Decisions API as OpenAI's answer to Jev. OpenAI's version runs as a hosted API on a specialized GPT-6 Luna and has no published price or accuracy figure yet, so a direct comparison needs your own tests.
What can the OpenAI Decisions API be used for?
OpenAI says developers supply context as text or images and get back answers they can use to classify content, route requests, or choose an agent's next action. A typical case is a support app deciding whether a message goes to technical support, sales, or complaints — the Decisions API returns the category directly, with a confidence score.

Sources · 4 outlets

Tags

  • openai
  • decisions-api
  • gpt-6-luna
  • classification
  • routing
  • decision-models
  • jev
  • devday
  • api
  • launch

← All releases · Learn AI