OpenAI's Decisions API is out of beta: fast yes-or-no answers for apps
The Decisions API gives apps a quick, typed answer instead of a chat reply: a probability, one choice from a list, or a score. OpenAI says it's about 10 times faster than its Responses API, and it charges only for input.
By ScoviAI Team
Key points
- The Decisions API is now generally available after a public beta
- It returns a probability, a choice or a score instead of free text
- OpenAI says it's about 10x faster than the Responses API
- $0.10 per million input tokens, $0.01 cached, and no output charge
OpenAI's Decisions API is now generally available. It's a new way for apps to ask an AI model a question and get back a ready-to-use answer, instead of a paragraph of text the app then has to interpret.
Think of the small judgment calls apps make all the time: Is this message spam? Which department should handle this ticket? How well does this answer match our guidelines?
What the Decisions API returns
You send text, images or JSON and a question. The API answers in one of three forms:
- Predicate: the probability that something is true, for example "this review contains a complaint."
- Choice: one value from a list you supply, such as "billing", "technical" or "sales."
- Score: a rating on a rubric you define.
Because the answer is already a number or a label, an app can act on it right away. OpenAI says the API is about 10 times faster than the Responses API for this kind of work.
What it costs
| Decisions API | Price |
|---|---|
| Input tokens | $0.10 per 1M tokens |
| Cached input tokens | $0.01 per 1M tokens |
| Output tokens | No charge |
Caching wasn't available during the beta, so repeated instructions are now much cheaper.
What's new since the beta
- Up to 400 questions per request, twice the earlier limit.
- Native JSON inputs and images inside question instructions.
- Conversation history, usage reporting for organizations, a link to OpenAI's voice model, and private network connections for companies.
Requests go to one endpoint and, for now, run on a single model, gpt-6-luna. Files and audio aren't supported as inputs yet.
What it means for you
- If you build apps, the Decisions API can replace prompts where you ask a chat model to "answer only yes or no" and then parse the reply.
- If you just use AI apps, you may notice some of them sorting, filtering and flagging things faster.
Sources
- Build more responsive apps with the Decisions API (OpenAI)
- Decisions guide (OpenAI developer docs)
ScoviAI Team is the ScoviAI newsroom. We write every article ourselves from the original announcements and link our sources. Spotted a mistake? Tell us and we'll fix it.
ADVERTISEMENT
