A Guth Labs publication ↗
Back to News

OpenAI adds discounted two-hour and eight-hour API windows

Green glass hourglass connected by a gold link to a layered glass and metal computing block
AI-generated editorial illustration

OpenAI added discounted two-hour and eight-hour completion windows to the Responses API in a public beta dated November 17 in its API changelog. Eligible model token rates are 30% below Standard for the two-hour option and 40% below Standard for the eight-hour option.

The beta supports GPT-6.2 Astra, GPT-6.2 Sol, GPT-6.1 Terra and GPT-6 Luna. According to the background-mode documentation, applications must use background execution and Standard processing, with streaming disabled or omitted.

The completion target begins when the API accepts the initial request and applies to that response, not an entire multi-step agent workflow. Eligible follow-up requests can also receive the discount. The default, soonest-possible background option remains available without these discounts.

Discounted requests share a queued-input-token limit with Batch API jobs. OpenAI also notes that background mode temporarily stores response data even when storage is disabled or a project uses Zero Data Retention.

The options give developers another cost-versus-waiting-time trade-off for work such as overnight document review, without requiring every step of a tool-using workflow to be submitted in advance.

Sources and citations

Revision history

Published .