stream disconnected before completion

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

• #37652
• #37980
• #37992

Powered by Codex Action

I am also facing this issue since morning.

Log: \[System Error\] stream disconnected before completion: 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 240ea33f-248a-4429-a163-acfcd317d05c in your message.

tream disconnected before completion: An error occurred while processing your request. You can retry your
😞

This is so frikin irritating please fix it asap since morning the progress is crap

This is so frikin irritating please fix it asap since morning the progress is crap

My friend is also facing same error and he might loose his job LOL

This is so frikin irritating please fix it asap since morning the progress is crap

BTW guys, I restarted my router lol and it works now fine. No error since then, did around 10 tasks till now.

what router? i started a new session and shifted context, worked fine for some time again started getting moody

what router? i started a new session and shifted context, worked fine for some time again started getting moody

i mean your network AP/Router xD

I reproduced the same 2026-08-11 incident on Arch Linux across both Codex CLI 0.147.0 and a third-party client using Codex app-server (T3 Code). This was not isolated to one frontend, one task, or a stale local process.

Feedback ID: 019ff0c9-9fbc-7532-a242-8ec6c3382682

Environment:

• Codex CLI 0.147.0, current npm install
• ChatGPT authentication
• gpt-5.6-sol at both xhigh and low effort
• old and brand-new sessions
• Arch Linux x86_64

Deduplicated result:

• 21 terminal stream disconnected before completion turn failures across five underlying sessions
• 11 additional request-ID failures in two retry loops immediately after a full machine reboot
• 32 unique OpenAI request IDs total
• WebSocket retries failed, then the automatic HTTPS fallback also failed
• structured disconnect records had httpStatusCode: null; no observed 401/403/429

The reboot control is useful: Wi-Fi was connected by 2026-08-11 14:26:16 UTC. The resumed CLI task failed repeatedly from 14:27:04 through 14:30:23 UTC, at both xhigh and low effort, with no NetworkManager disconnect. Those turns were interrupted after 152 s and 72 s because retries continued.

Local health controls are green: codex doctor reports 0 failed checks, a successful HTTP 101 WebSocket handshake, reachable provider endpoints, valid ChatGPT auth, and consistent/current install state. A fresh task also failed, so this was not context-window exhaustion.

The service later recovered. Controlled exact-response probes succeeded in the same old/new T3 and CLI sessions at 14:59–15:03 UTC, including the original troubleshooting session. That makes this look like an intermittent backend/request-routing incident rather than a persistent local network or install problem.

This also matches the same-window CLI report in #38001. I can provide the two sanitized JSON request-ID inventories if maintainers want them.

<details>
<summary>32 request IDs and timestamps</summary>

Terminal turn failures:

• f2a90fe1-0bad-4f4e-9146-3292a688e4aa — 07:50:43 UTC — app-server client
• d338310a-3fd5-4cd7-b515-b8c74be9905c — 12:12:07 UTC — app-server client, new task
• c1fe7992-bafb-41e7-9f13-40e3b273b068 — 12:12:35 UTC — app-server client, old task
• 54bd1e18-9366-4101-828b-f604fdbba46c — 12:16:24 UTC — direct CLI, fresh task
• 1112e78b-7651-4f30-88b7-efc113c17f8d — 12:26:18 UTC — app-server client, old task
• d87c85fa-5c09-45d1-9386-fbec08558bad — 12:33:13 UTC — direct CLI
• c1a2e5f8-08d8-42ac-8311-f447000991a2 — 12:37:59 UTC — direct CLI
• 70eec2d2-3e14-4827-b7c7-c1bdc77b0dfa — 12:40:05 UTC — direct CLI
• 19af161d-dcfb-4105-beec-b90e8647087d — 12:41:48 UTC — direct CLI
• b0a18a2e-faaa-4f08-b75a-0ac342ff85e0 — 12:45:12 UTC — direct CLI
• fd609cd0-8bd1-4bb0-a321-adbb57073c5d — 12:52:10 UTC — direct CLI
• cb5f1567-7adb-47a3-aca8-2915f8578fd8 — 12:57:14 UTC — direct CLI
• 4d54d374-8ba5-4aa3-a898-3a6f41b8370b — 12:59:04 UTC — direct CLI
• 95bb59c4-98c1-4716-859c-68e0e8bc5f46 — 13:04:23 UTC — direct CLI
• ed2bdc51-d8b6-47e1-a3ef-3485a2381e49 — 13:07:46 UTC — app-server client, old task
• c09969b6-bd82-4e2f-b16d-4f7bafba47e8 — 13:18:39 UTC — direct CLI
• ee1d3d0c-e509-489e-9ada-3c1bd3eddd77 — 13:56:57 UTC — direct CLI
• 102d5525-0b7f-4919-8e82-64c50c294909 — 14:04:54 UTC — direct CLI
• 9907973e-2ebd-4d97-905f-cdb5b1bd07ba — 14:11:09 UTC — direct CLI
• de5143b9-52e4-4b25-a26a-cc6937031ce7 — 14:17:35 UTC — direct CLI
• 0e745566-0262-401a-ba94-eb311c4d74fb — 14:23:51 UTC — direct CLI

Post-reboot retry failures:

• a67507f0-d1c0-4892-83b9-1dc6045a775c — 14:27:04 UTC — xhigh
• 322ab22a-d084-4bf3-bacc-68f293e7a740 — 14:27:27 UTC — xhigh
• 16c1dcd1-9b5e-40b7-a223-57e146ec8ca8 — 14:27:54 UTC — xhigh
• bbe1108f-9372-4c2e-8215-1d33b5f6fb04 — 14:28:14 UTC — xhigh
• 7874ef55-b096-466a-80db-164c4209e330 — 14:28:30 UTC — xhigh
• b48d79a7-ceed-4143-9608-1b7d3ffcbe6f — 14:29:13 UTC — xhigh
• cdeab1ad-f7ed-42cb-b72b-748fe1eff4d4 — 14:29:37 UTC — low
• c4230363-b6cc-44a1-9aa1-dd9eedc346ac — 14:29:46 UTC — low
• a0906f8a-761e-4060-97f9-8a277e1aafc9 — 14:29:52 UTC — low
• 6a8b0fe5-b106-4e7b-a758-6da69839ca0b — 14:29:56 UTC — low
• 5ed54e38-ecfd-47d8-8256-8b7767531096 — 14:30:23 UTC — low

</details>

This happens to me in just about every session. Sitting there waiting 45 minutes for something I could have done myself in a couple of minutes. It doesn't seem worth making a bug report/issue here because OpenAI doesn't seem to respond to these issues, what's the point?