Seeing a spike in failures from the Responses API today (2026-05-14) when using background=true. Symptoms:
- Stream initiates successfully, response is created and starts processing
- Mid-stream, the connection terminates with the generic error: “An error occurred while processing your request. You can retry your request, or contact us through our help center at help.openai.com if the error persists. Please include the request ID wfr_… in your message.”
- Request IDs carry the wfr_ prefix, suggesting failures inside the background workflow/queue layer rather than the synchronous API path
- Reproducible across two independent deployment environments hitting the same OpenAI org, so unlikely to be client-side
- Our app-side retry logic (5x initial create, 4x mid-stream, exponential backoff capped at 30s — total ~60s window) is exhausting before recovery, so whatever is happening on OpenAI’s side persists > 60s per turn
What do i do?