OpenAI has introduced a "completion window" option to the Responses API background mode, enabling long-running tasks to execute for up to 8 hours while offering significant token cost reductions. When users select the 2-hour or 8-hour completion windows, model token rates drop by 30% and 40%, respectively, compared to Standard processing. The default soonest window provides no discount. The feature is currently in public beta on GPT-6.2 Astra, GPT-6.2 Sol, GPT-6.1 Terra, and GPT-6 Luna. Discounted windows require Standard processing, do not support streaming output, and share rate limits with the Batch API. Subsequent requests using the same continuation_id receive higher queue priority. OpenAI also disclosed that starting with GPT-6.2 Astra and Sol, new models no longer offer Flex processing.