Glonce

OpenAI launches Decisions API beta for yes/no and pick-one calls

productsThe Glonce editorsThe Decoder🔥59Glonce rating59Impact45Novelty55Relevance95Surprise30Players75Credibility50
OpenAI launches Decisions API beta for yes/no and pick-one calls

OpenAI announced two updates to its API platform, according to The Decoder.

The first is a new Decisions API, which evaluates text, images, or both. It is in public beta now, with general availability described as coming soon. The article says it runs about ten times faster than the Responses API, without giving a benchmark or latency figure. So far only the gpt-6-luna model is supported, priced at $0.10 per million input tokens. Output tokens are free.

The API handles three response types: yes/no probabilities, picks from predefined categories, and scale-based ratings. Use cases named, attributed to OpenAI, are damage detection in photos, automatic customer inquiry routing, and document classification. The API supports zero-data retention and HIPAA-compliant use in the US and Europe.

The Decoder adds its own read: the Decisions API is likely OpenAI's response to a mini-trend of "decision models" that Jev kicked off in mid-September. That is the outlet's assessment, not an OpenAI statement.

The second update is a consolidation of paid API tiers. OpenAI cut them from five down to three: Build, Launch, and Grow. Organizations move up automatically when their total credit purchases hit the next threshold. Monthly usage limits sit at $500, $5,000, and $200,000.

Key facts

  • OpenAI's new Decisions API is in public beta and evaluates text, images, or both; general availability is described as coming soon.
  • It returns one of three response types: yes/no probabilities, picks from predefined categories, or scale-based ratings.
  • Only gpt-6-luna is supported so far, at $0.10 per million input tokens, with free output tokens; the article says it runs about ten times faster than the Responses API.
  • It supports zero-data retention and HIPAA-compliant use in the US and Europe.
  • OpenAI also cut its paid API tiers from five to three (Build, Launch, Grow), with monthly usage limits of $500, $5,000, and $200,000.

Why it matters

The Decisions API narrows what a model returns to a yes/no probability, a category or a rating, rather than open-ended text. That is the shape of many routine classification jobs, and OpenAI pairs it with a low input price and free output tokens. The Decoder reads the launch as likely a response to a mini-trend of "decision models" that Jev kicked off in mid-September; that is the outlet's speculation, not OpenAI's stated motive. The tier cut from five to three is a smaller change to how API access is organised.

Who it affects

The use cases OpenAI names are damage detection in photos, automatic customer inquiry routing, and document classification, so teams building triage, intake or sorting features are the obvious audience. The zero-data retention and HIPAA-compliant options in the US and Europe are aimed at organisations with sensitive data. Anyone on the paid API tiers is affected by the consolidation to Build, Launch and Grow.

How to use it

The Decisions API is available now in public beta, and only gpt-6-luna works with it for the moment. Pricing is $0.10 per million input tokens, and output tokens are free. You choose among three response types: yes/no probabilities, picks from predefined categories, or scale-based ratings. On tiers, organizations move up automatically when their total credit purchases hit the next threshold. Monthly usage limits sit at $500, $5,000, and $200,000.

How solid is it

The facts come from a single report by The Decoder, which describes OpenAI's announcement. The use cases are attributed to OpenAI. The speed claim, about ten times faster than the Responses API, comes with no benchmark, latency figure or test conditions. No accuracy figures or quality comparisons for the Decisions API are given, and no customer names. The model name gpt-6-luna is given only as the single supported model.

Risks and caveats

This is a beta, and general availability is only described as coming soon, with no date. Pricing is given for gpt-6-luna alone; none is given for other models or for the Responses API. The source does not say which usage limit belongs to which tier, what the credit-purchase thresholds are, whether the tier cut changes pricing, or what the old five tiers were. The link to the mid-September "decision models" trend is The Decoder's own inference.