OpenAI Launches Decisions API with Enhanced Features

Decisions API Now Generally Available

OpenAI announced on October 22, 2026, that the Decisions API is now generally available. Built with GPT-6 Luna, this service is designed to handle bounded-answer questions, enabling applications to make near real-time decisions up to 10 times faster than using the Responses API.

Key Features and Capabilities

The Decisions API accepts context in the form of text, images, or native JSON. It supports up to 400 questions per request, which is double the previous limit. Developers can define a question along with a set of allowed answers, and the API returns one of three answer types:

  • Probability: An estimate from 0 to 1 indicating the likelihood that a condition is true.
  • Choice: One of the predefined options, along with associated probabilities and confidence levels.
  • Score: A probability-weighted average of level indices in a rubric.

Integration with GPT-Live

Since the public beta, OpenAI has added a native integration with GPT-Live, which enhances the API’s ability to make low-latency decisions during live conversations. This integration allows applications to respond quickly and effectively, improving user interactions.

Performance and Use Cases

OpenAI has reported various internal and customer use cases across product and business workflows. Examples include classifying transaction descriptions, routing document processing, and enhancing language-learning experiences. Notably, Box’s VP of Product Management, Biswaroop Palit, stated that their evaluation found the Decisions API delivered approximately 5 times faster route selection on 30 certificate of insurance cases while maintaining the accuracy of the Responses API.

Pricing and Availability

The Decisions API is available via a dedicated endpoint, with regional processing options in the US and EU. Eligible customers can access Zero Data Retention features. Pricing is set at $0.10 per million input tokens, with cached input billed at $0.01 per million tokens for supported cache durations, including 30-minute and 12-hour cached writes. Importantly, only input tokens incur charges, with no fees for output tokens.

Developer Resources

OpenAI provides developer tools such as a Decisions Playground and a comprehensive developer guide to facilitate testing with text or images. This support aims to help developers effectively implement the Decisions API in their applications.


Original source: OpenAI News

Leave a Comment