DEV Community

DebugRadar
DebugRadar

Posted on Originally published at debugradar.com

Fix: ChatGPT Work Mode HTTP 500 Error

#ai #chatgpt #openai #troubleshooting

Originally published on DebugRadar.

ChatGPT Work Mode is returning HTTP 500 on every message, including brand-new conversations — and as of November 12, 2026, it has been broken for over 38 hours with no improvement. OpenAI Support confirmed the failure is server-side and escalated it for technical investigation. Then Support went quiet for 36 hours. Regular Chat works fine on the same account. This is not a problem you can fix from your side — but you can confirm that in two minutes and keep working today.

Quick Fix Summary

  • Problem: ChatGPT Work Mode returns HTTP 500 on POST /backend-api/f/conversation for every message, including brand-new conversations, while regular Chat works normally.
  • Fix: There is no user-side fix. OpenAI Support confirmed a server-side failure and escalated it, with no timeline and no reply in 36 hours.
  • Workaround: Use regular Chat or Codex Work for the same tasks, and send OpenAI Support your request IDs, timestamps, and a HAR file.

Technical Root Cause

Error Code Affected Version Root Cause Current Status
HTTP 500 on POST /backend-api/f/conversation ChatGPT Plus, Work Mode, model GPT-6.1 Sol (Low), all devices (iPhone, Safari, desktop Chrome on Windows) Server-side failure during Work's internal preparation, before the response begins — confirmed by OpenAI Support from internal logs Still broken as of November 12, 2026 — 38+ hours; OpenAI's status page shows Work as Operational; no fix or timeline; Support has not replied in 36 hours

Step-by-Step: Confirm It and Keep Working

Step 1 — Confirm the Failure Is on OpenAI's Side

  1. Open status.openai.com and check the ChatGPT components. The status page shows Work as Operational, which matches what users see: this hits specific accounts and workspaces, not every user at once.
  2. Send a message in a brand-new Work conversation. If it fails with "Network Error" or "Something went wrong", the problem is not your chat history.
  3. Send the same message in regular Chat. If regular Chat answers and Work fails on the same account, the failure is not your browser, network, or plan.
  4. Try another device or network — Incognito mode, another browser, mobile data. Users who tried all of these got the same HTTP 500. If Work fails everywhere and regular Chat works everywhere, it is server-side.

Step 2 — Capture Evidence for OpenAI Support

If you have not contacted Support yet, send them what they ask for — it is the only thing that moves a case like this:

  1. Open Chrome DevTools (F12 → Network tab) and reproduce the failure in Work Mode.
  2. Find the failed request: POST /backend-api/f/conversation → HTTP 500. Copy its request ID.
  3. Export a HAR file: in the Network tab, right-click and choose Save all as HAR with content. Remove any personal data from it before sending.
  4. Send Support the request IDs, the timestamps of the failed attempts, the conversation URLs, and the sanitized HAR file.

The user who reported this on November 10 sent all of it 24 hours ago and has heard nothing back since — set your expectations there.

While Work Mode is down, keep working
A broken Work Mode should not stop your workday. Claude Pro gives you a full AI workspace to keep going in — projects, files, and long tasks — while OpenAI sorts out Work Mode. Gemini Advanced is the other fallback worth having open in a second tab. (Affiliate link — DebugRadar earns a commission if you sign up through it.)

Step 3 — Work Around It Today

Use regular Chat or Codex Work for the same tasks. Both run normally on the same account while Work Mode fails. Move the files and context you need into a regular chat or a Codex session and carry on — do not wait on Work Mode to come back before doing the work.

Step 4 — What NOT to Do

  • Do not clear all your browser data, reinstall the app, or reset your account hoping for a fix. Users tried all of it across devices and networks. The failure is server-side.
  • Do not pay for a third-party "fix" tool. No tool can fix an HTTP 500 that comes from OpenAI's servers.
  • Do not hammer the retry button. Retrying hard during a backend failure can trigger rate-limit messages on top of the 500s.

FAQ

Is ChatGPT Work Mode down for everyone?
No. OpenAI's status page shows Work as Operational. The failure hits specific accounts and workspaces — regular Chat works on the same account. That is why it persists while no global outage is declared.

Will clearing my cache or reinstalling the app fix it?
No. Users tried Incognito mode, other browsers, other devices, and mobile data — same HTTP 500. The failure happens during Work's preparation step on OpenAI's servers.

When will OpenAI fix it?
No timeline exists. Support confirmed the server-side failure and escalated it, then went quiet for 36 hours. Keep working in regular Chat or Codex Work, and check the community thread for updates from other affected users.

Sources: OpenAI Developer Community thread, updated November 12, 2026 — "It's now November 12, and the problem has persisted for over 38 hours without any improvement." (https://community.openai.com/t/chatgpt-work-mode-completely-stopped-working-http-500-across-all-devices-anyone-else-experiencing-this/1415391). Related thread: "ChatGPT Work broken for over 48 hours — server-side failure confirmed, but OpenAI Support has stopped responding" (https://community.openai.com/t/chatgpt-work-broken-for-over-48-hours-server-side-failure-confirmed-but-openai-support-has-stopped-responding/1412217).

Read the full guide on DebugRadar: Fix: ChatGPT Work Mode HTTP 500 Error

Top comments (0)