Source: Background mode (OpenAI guide)
OpenAI's guide to running long Responses API calls asynchronously with background: true, polling, cancellation and resumable streaming.
Contents
- What: official API documentation page
- Publisher: OpenAI
- Date: living document; read 2026-11-14
- Original: Background mode
Summary
How to start long-running reasoning or agentic responses without holding a connection open, then poll, cancel or resume streaming.
Key claims
background: trueruns the response asynchronously; poll GET while status isqueuedorin_progress; other statuses are terminal.POST /responses/{id}/cancelstops it and is idempotent.- With
backgroundandstreamboth true, track each event’ssequence_number; after a drop, resume withstream=trueandstarting_after. - Time to first token is higher than for synchronous calls.
- Background mode stores response data temporarily even with
store=false; for ZDR projects it is kept for about 10 minutes.