OpenAI Launches Decisions API Powered by GPT-6-Luna

OpenAI has introduced the Decisions API, a new tool designed to streamline complex evaluations of text, images, or multimodal inputs. Currently available in public beta with general availability expected soon, the API is engineered to run approximately ten times faster than the standard Responses API. At launch, the system exclusively supports OpenAI's gpt-6-luna model, offering an incredibly low price point of $0.10 per million input tokens, while making output tokens entirely free.

The new API simplifies outputs into three distinct formats: yes-or-no probabilities, selections from predefined categories, and scale-based ratings. According to OpenAI, these structured outputs are ideal for automated workflows such as identifying damage in photographs, routing customer inquiries to the correct departments, and classifying large volumes of documents. To accommodate enterprise needs, the Decisions API supports zero-data retention and is HIPAA-compliant for users in both the United States and Europe.

This release appears to be OpenAI's entry into the emerging category of "decision models," a trend initiated by Jev in mid-September. For developers and system architects, the combination of high-speed processing and free output tokens significantly lowers the financial and computational barriers to deploying high-volume classification pipelines.

Alongside the new API, OpenAI simplified its paid developer tiers, reducing them from five levels down to three: Build, Launch, and Grow. Under this new structure, organizations automatically advance to higher tiers as their total credit purchases reach the required thresholds. The monthly spending limits for these consolidated tiers are now set at $500 for Build, $5,000 for Launch, and $200,000 for Grow.