OpenAI makes Decisions API generally available at $0.10 per million input tokens
OpenAI promoted its Decisions API to general availability, built on GPT-6 Luna, supporting up to 400 questions per request, with no separate output charge and cached input at $0.01 per million tokens, up to 10x faster than the Responses API.
OpenAI announced on October 22 that the Decisions API is generally available, ending its research preview. The interface targets applications that need a structured pick from a set of options, with the key difference from a general chat interface being that the output can be consumed directly as a decision.
On pricing, input is $0.10 per million tokens, cached input is $0.01 per million tokens, and there is no separate output charge. OpenAI says this delivers up to a 10x latency improvement over running equivalent tasks through the Responses API. A single request accepts up to 400 questions for bulk evaluation.
The API is built on GPT-6 Luna and integrates GPT-Live natively so the decision step can reach live information within the same chain. OpenAI also noted US and EU data residency options so enterprise customers can select where data lands by region.