Sponsored segment: an API for quick decisions (predicate probability, choice from options, score ranking), up to 10x faster than the Responses API in GPT-6 Luna, at 10 cents per million uncached input tokens with no output charges.
OpenAI API that returns bounded decisions such as probabilities, choices from a set, or scores instead of free text.
Updated EU status checked Last shown
OpenAI Decisions API at a glance
- What is OpenAI Decisions API?
- OpenAI Decisions API is an api from OpenAI. OpenAI API that returns bounded decisions such as probabilities, choices from a set, or scores instead of free text.
- Who makes OpenAI Decisions API?
- OpenAI Decisions API is made by OpenAI.
- Does OpenAI Decisions API work in the EU?
- Available in the EU. The OpenAI API is offered in a supported-countries list (not shown in full here) and regional processing (data residency) endpoints exist at a 10% uplift. No explicit EU exclusion for the Decisions API was found, but its specific EU data residency support is not confirmed. Last verified 14 Nov 2026.
- How much does OpenAI Decisions API cost?
- API pricing. Per creator reports: $0.10 per million uncached input tokens, no output charges. Regional processing endpoints carry a 10% uplift for models released on or after March 5, 2026. Not confirmed on the provided pricing page excerpt. Licence: Proprietary.
- Where was OpenAI Decisions API shown?
- In 5 creator videos by Matt Wolfe, Matthew Berman. Most recently by Matt Wolfe on 13 Nov 2026, at 9:50 in “AI News: Things Will Never Be The Same”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
The Decisions API is an OpenAI developer API for fast, constrained model outputs. Rather than generating free-form text, it returns bounded answers: the probability that a predicate is true, a choice from a given set of options, or a score or ranking. According to creator coverage, it is built on GPT-6 Luna and is now generally available.
It is aimed at classification, routing, moderation-style checks and ranking tasks where latency and cost matter more than prose output. A sponsored segment described it as up to 10x faster than the Responses API with GPT-6 Luna. The same segment described pricing as $0.10 per million uncached input tokens with no output token charges. These figures come from creator reports and were not confirmed on the pricing page excerpt provided.
It is accessed through the OpenAI API platform and documented in the developer guides. Because it returns only bounded values, it is not suited to open-ended generation. Regional processing (data residency) endpoints carry a 10% uplift for models released on or after March 5, 2026.
- Returns bounded outputs rather than free text
- Predicate probability: likelihood that a statement is true
- Choice from a provided set of options
- Score or ranking outputs
- Built on GPT-6 Luna; reported up to 10x faster than the Responses API
- Reported pricing: $0.10 per 1M uncached input tokens, no output charges
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Now generally available, built with GPT-6 Luna; returns bounded answers such as probabilities, choices from a set, or scores rather than free text.
API announced at Dev Day that returns predefined decisions (predicates, choices, scores) instead of long text. It uses GPT-6 Luna and is meant to be ultra-fast and cheap.
API for real-time decision-making that outputs one of finite predefined answers to user-defined questions instead of text. Creator sees it as OpenAI's answer to Jev from the previous week.
Preview API of a model built for quick decisions, based on Luna, OpenAI's lightest and smallest model. The creator thinks a purpose-built decision model like Jev would still be faster.
Similar AI tools
Closest neighbours by meaning, not just by category.
Gemini Decisions API
API · Google
Google's API for AI-driven decision-making, reported by creators as a counterpart to similar OpenAI and Microsoft offerings.
GPT-5.6 Luna
Model · OpenAI
OpenAI's lighter, lower-cost GPT-5.6 model variant, suited to routine tasks such as setup and scripting work.
OpenAI o1
Model · OpenAI
OpenAI's first reasoning model series, trained with reinforcement learning to think before answering; now deprecated in the API.
GPT-5.5
Model · OpenAI
OpenAI language model available through ChatGPT, Codex and the API, used for coding, review and general tasks.
GPT-6.1
Model · OpenAI
OpenAI language model in the GPT-6.1 family; the Soul Max variant is noted by reviewers for low token use per task.
GPT-5.4
Model · OpenAI
OpenAI frontier language model used for coding, reasoning and agentic tasks, available via ChatGPT and the API.