IT之家 · 智能时代 · 10/23/2026, 07:48:13
OpenAI Launches Decisions API on GPT-6 Luna with $0.1/M Input Pricing
OpenAI has fully launched the Decisions API, built on GPT-6 Luna and optimized for bounded-answer queries to enable near-real-time agent decisions. The API offers up to 10x faster response times compared to the standard Responses API and features an aggressive pricing model charging only for input tokens ($0.1 per million), while supporting native JSON inputs, image instructions, and prompt caching.
SOURCE COVERAGEFull Original Report
On October 23, it was reported that on October 22 (local time), OpenAI announced the general availability of its Decisions API.
According to the introduction, the API is built on GPT-6 Luna, focusing Luna’s model intelligence on questions with bounded answers to help applications and agents make near-real-time decisions based on text, images, and structured inputs — delivering response speeds up to 10 times faster than calling GPT-6 Luna via the Responses API.
Regarding pricing, the Decisions API charges 0.01 per million cached input tokens (approximately ¥0.067 RMB). The API bills only for input tokens; output tokens are free. Across OpenAI, teams are applying the Decisions API to product and business workflows.
OpenAI stated that since the public beta, the feature has added native GPT-Live integration, prompt caching for reused contexts, and support for inference data residency in the United States and the European Union.
Additionally, developers can now pass native JSON inputs, include images within question instructions, and submit up to 400 questions in a single request (double the previous limit). OpenAI noted ongoing optimizations in latency performance, including expanded service capacity across multiple regions in Asia.