OpenAI launches decision API with gpt-6 luna for real-time insights

On October 22, 2026, OpenAI announced the full release of its Decision API, built on the GPT-6 Luna model. This API is designed to focus on questions with bounded answers, enabling applications and agents to make near real-time decisions based on text, images, and structured inputs. The response speed is up to 10 times faster than using the Responses API with GPT-6 Luna. The Decision API charges $0.10 per million input tokens, with $0.01 per million for cached inputs. Output tokens are free. OpenAI also introduced native GPT-Live integration, prompt caching for reused contexts, and data residency support in the U.S. and EU. Developers can now submit up to 400 questions per request, double the previous limit, and include native JSON inputs and images. OpenAI expanded service capacity in Asia to improve latency.