OpenAI Launches GPT-6 Luna-Powered Decisions API for General Availability

OpenAI has launched the Decisions API, powered by GPT-6 Luna, to accelerate AI decision-making processes. The new application can answer limited-response queries submitted via text, image, and structured JSON inputs up to 10 times faster than using GPT-6 Luna via the Responses API.

Advanced Features and Response Types

Updated following its public beta, the Decisions API works in direct integration with GPT-Live for real-time voice and chat-focused decision-making processes. The system offers prompt caching support for reused contexts, data processing location guarantees in the US and the European Union, and expanded capacity across Asia. Supporting native JSON inputs and visual contexts, the application allows for querying up to 400 questions in a single request.

The service supports three core decision structures: Predicate, which estimates the probability of a specific condition occurring between 0 and 1; Choice, which selects from provided options and returns probability and confidence values; and Score, which evaluates inputs based on specific criteria and returns a weighted average index.

Use Cases and Cost Structure

In addition to product, finance, and communications teams within OpenAI, external developers are also utilizing the API for areas such as generating interactive user responses, financial transaction categorization, and narrative scope classification.

Decisions API pricing is based on input tokens. A fee of $0.10 per million input tokens has been set, with $0.01 per million for cached tokens, while no additional fees are charged for output tokens. For eligible customers, Zero Data Retention options are provided alongside regional data processing capabilities in the US and EU.