OpenAI has officially launched its Decisions API, making the tool available to all users with API access as of October 22, 2026. The service is priced at $0.10 per million input tokens, with billing based exclusively on input volume.
Powered by the GPT-6 Luna model, the API is engineered for high-volume, low-latency tasks such as request routing and input classification. It delivers structured, typed outputs including probabilities, scores, and specific choices rather than long-form text.
The platform now supports up to 400 questions per request, an increase from the capacity offered during the beta phase. OpenAI reports that the API operates up to 10 times faster than the standard Responses API for bounded decision tasks, with demonstrations showing latency as low as 150 milliseconds.
Updates introduced since the beta period include native GPT-Live integration, prompt caching, and support for JSON and image inputs. The API is HIPAA eligible and supports Zero Data Retention, targeting markets in both the US and Europe.
