stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses) #12207

Activity

1. bsheoran-chwy

   added

   appIssues related to the Codex desktop appIssues related to the Codex desktop app

   on Feb 19, 2026
2. github-actions

   added

   bugSomething isn't workingSomething isn't working

   codex-webIssues related to Codex WebIssues related to Codex Web

   on Feb 19, 2026
3. github-actions commented on Feb 19, 2026

   

   github-actionsbot

   on Feb 19, 2026 – with GitHub Actions

   Contributor

   More actions

   Potential duplicates detected. Please review them and close your issue if it is a duplicate.

   • stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses) #12206
   • stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses) #10985
   • stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses) #11741
   • Stream disconnected before completion on VPS using ChatGPT OAuth (worked 12 hours prior, reproducible, network verified) #11743
   • thread ID 019c4d06-0a03-7ba0-b41d-68f0633d3474 #11453

   Powered by Codex Action
4. Nadeesha-D-Shalom commented on Feb 19, 2026

   

   Nadeesha-D-Shalom

   on Feb 19, 2026

   More actions

   I’m investigating this behavior as well and wanted to share some structured observations.

   This error pattern:

   "stream disconnected before completion: error sending request for url (https://chatgpt.com/backend-api/codex/responses)"

   suggests the client streaming connection (likely SSE or HTTP/2) is being interrupted before the backend completes the response.

   Some possible technical causes worth checking:

   1. HTTP/2 stream reset (RST_STREAM) or idle timeout on macOS network stack
   2. Enterprise OAuth token refresh occurring mid-stream
   3. Gateway/proxy terminating long-lived streaming connections
   4. Backend timeout for long-running responses
   5. Network layer TLS renegotiation failure

   Questions that may help narrow it down:

   • Does this only occur for longer prompts?
   • Does it happen after the session has been idle?
   • Are there any HTTP status codes logged before disconnect?
   • Does disabling VPN/firewall change behavior?
   • Is it reproducible on another network?

   If this is a streaming timeout issue, implementing automatic stream reconnection or partial-response recovery might mitigate the user-facing failure.

   Happy to help test further if needed.
5. etraut-openai

   removed

   codex-webIssues related to Codex WebIssues related to Codex Web

   on Feb 24, 2026
6. etraut-openai commented on Feb 24, 2026

   

   etraut-openai

   on Feb 24, 2026

   Contributor

   More actions

   Thanks for reporting the problem, but there's not enough information for us to investigate further. If you see this again, please use the /feedback slash command to upload your logs and session details and include the thread ID in the bug report.

   This error can be caused by may different things, and without logs it's difficult to say which of those you were experiencing.

   👎React with 👎1Reacted by Mathavan Venugopal
7. etraut-openai

   closed this as completedon Feb 24, 2026
8. wirthual commented on Feb 25, 2026

   

   wirthual

   on Feb 25, 2026

   More actions

   Using the vscode plugin I saw this error when using the API key to authenticate, using the login method worked for me.
9. smbell1979 commented on Sep 10, 2026

   

   smbell1979

   on Sep 10, 2026

   More actions

   This has been happening to me with Astra and making it 100% unusable.