200 OK from /v1/responses but body times out (UND_ERR_BODY_TIMEOUT)

API Bugs
,
21

@stevecoffey, any updates?

22

@stevecoffey yes, this is indeed needed. There are too many execution errors in the agentic run.

1 Like
23

I’ve got the same error(UND_ERR_BODY_TIMEOUT) on gpt-5.2-pro with reasoning_effort: xhigh. Test it twice, same error. When change reasoning_effort to high - all is working fine. Also on gpt-5.2 with xhigh - all fine too.

NodeJS openai lib

1 Like
24

@stevecoffey , any updates?

25

Hey all! I sincerely apologize for the delay, there was an issue preventing us from rolling this out over the last few days. We’re resuming the rollout of this feature now, expect it to be fully rolled out within the next 4 hours

4 Likes
26

In what form will this take?

  • Identified as an OpenAI server issue, terminating prematurely yet Status 200? or
  • an enhancement to the stream content, consumable or generally passive to clients?
27

we emit a keepalive streaming event every 30s to keep intermediate proxies and load balancers from terminating the connection if the gap between streaming events exceeds this deadline. This can happen, for example, if you are not using reasoning summaries with high reasoning efforts

5 Likes
28

Hey Everyone, Please let us know, If the issue still persists. Thank you!

1 Like
29

Although the frequency has decreased, we are still having terminated error with OpenAI Agents SDK which I believe is related to this issue.

30

Closing this thread and the original issue is fixed. baki.kucukcakiroglu we will continue the discussion in your thread. Thank you for flagging.

3 Likes