AnswersPublished18 sources
OpenAI’s Decisions API Goes GA: Features, Pricing, and Limits
OpenAI made the GPT 6 Luna powered Decisions API generally available on October 22, 2026. The listed input price is $0.10 per million tokens; launch coverage reports $0.01 per million cached input tokens and no output token charge.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- OpenAI made the GPT 6 Luna powered Decisions API generally available on October 22, 2026.
- The listed input price is $0.10 per million tokens; launch coverage reports $0.01 per million cached input tokens and no output token charge.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What changed when OpenAI made its GPT-6 Luna–based Decisions API generally available to all customers on October 22, 2026, following its Oct. Article summary: OpenAI’s October 22, 2026 announcement made the GPT-6 Luna–based Decisions API generally available, extending its fast, bounded-answer approach rather than introducing a conventional text-generation endpoint. The availab. Topic tags: general, documentation, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fak
OpenAI moved its GPT-6 Luna-powered Decisions API from public beta to general availability on October 22, 2026, following the October 6 beta launch. The endpoint is designed for application decisions with a bounded set of possible answers—not for generating a full prose response—and its new release features include native JSON input, image support in question instructions, prompt caching, and a limit of 400 questions per request. 2
4
5
How the Decisions API works
Developers provide a question with finite, predefined answers and context. The API returns a typed result—such as a probability, a choice, or a score—that software can use to classify content, route a request, or select an agent’s next action. OpenAI describes the approach as a way to make near-real-time decisions without generating a full model response. 4
11
13
The API accepts text and images, and the GA changelog adds direct structured-data input through input_json. It also allows images in individual question instructions. The request limit rose from 200 questions during beta to 400 at GA. 2
4
What changed at general availability
The documented GA additions include shared-input prompt caching for discounts, image support inside question instructions, and native JSON input. 2 OpenAI’s developer-community announcement also lists native GPT-Live integration, intended to let voice applications react while a person is speaking.
Launch coverage additionally reported inference data residency options in the United States and European Union. The available changelog excerpt does not list that detail, so it is best treated as a reported launch feature rather than a confirmed item in the changelog. 9
Speed: a claim about OpenAI’s own API
OpenAI says Decisions can return answers up to 10 times faster than using GPT-6 Luna through the Responses API. That is a comparison between two ways of using OpenAI’s model, not evidence that Decisions is faster than Jev or other decision models. 4
The release materials supplied here do not provide a matched, independently verified benchmark across competing APIs. They also do not establish the proposed image-calibration advantage or roughly 60-millisecond video-frame decisions, so those should not be treated as demonstrated differentiators.
Pricing and customer results
The OpenAI guide excerpt lists regular input at $0.10 per million tokens. Launch coverage reports $0.01 per million cached input tokens and no charge for output tokens; the GA changelog confirms prompt-cache discounts but does not state the exact cached rate in its excerpt. Check the current pricing documentation before estimating production costs. 1
2
9
Coverage identifies Box, Datadog, Ramp, and Speak among companies that tried the API, with engineers using it to replace some existing large-model calls to increase throughput. The available material does not give comparable, quantified performance results for Box, Datadog, or Speak, so it does not support a reliable company-by-company claim about speed, accuracy, or cost savings.
How it compares with Jev—and what remains unknown
TypeSafe’s Jev helped establish the decision-model category, according to Forbes coverage, and that report lists Jev’s starting price at $0.042 per million input tokens, with no output-token charge. On those reported list prices, Jev’s input rate is lower than the Decisions API’s stated $0.10 rate; actual cost still depends on token use and caching. 1
TechCrunch reported that OpenAI’s team adapted GPT-6 Luna to speed up inference and lower costs rather than building a specifically calibrated decision model. That is a reported account of the product’s development, not a detailed technical disclosure from OpenAI. The same outlet reported TypeSafe’s valuation at $7.6 billion; that figure should likewise be understood as a reported valuation, not independently established here. 10
The available sources do not provide enough comparable evidence to rank Decisions against Jev, Liquid’s D1 Ultra, or Base Lab’s Elephant-1 on the AIM System One or Artificial Analysis Decision indexes. They also do not verify Jev’s alleged text-only limitation or its exact OpenRouter popularity. The practical takeaway is narrower: Decisions is now generally available as a multimodal, bounded-answer API, with a larger request limit and a speed claim measured against another OpenAI endpoint—not proof of benchmark leadership across the category. 2
4
10
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "OpenAI’s Decisions API Goes GA: Features, Pricing, and Limits"?
OpenAI made the GPT 6 Luna powered Decisions API generally available on October 22, 2026.
What are the key points to validate first?
OpenAI made the GPT 6 Luna powered Decisions API generally available on October 22, 2026. The listed input price is $0.10 per million tokens; launch coverage reports $0.01 per million cached input tokens and no output token charge.