OpenAI Decisions API — GPT-6 Luna (GA) — Event History

Official 2026-10-22 GA of /v1/decisions using gpt-6-luna: 400 typed questions/request (was 200 beta), shared-input prompt caching, question-specific images, input_json and usage reporting. Base uncached input remains USD 0.10/1M and output remains free; cache reads USD 0.01/1M and cache writes billed at standard Luna rate USD 0.125/1M. This is an endpoint capability and billing-terms change, not a same-context uncached native-price rise.

New public beta Decisions endpoint uses gpt-6-luna at USD 0.10 per 1M input tokens; no output-token charges. Regional and long-context premiums apply.