Surface: Web / iOS / Android / macOS / Windows / Linux
I have seen the same issue with varying UIs in Web, iOS and macOS.
Browser or app version:
Arc Version 1.165.0 (87194) Chromium Engine Version 153.0.8010.48
App version 26.917.71314
Affected feature and model, if relevant:
ChatGPT Work in the Cloud
Bug
What happened, and what did you expect?
Re-opening an previously created ChatGPT Work thread created either on chatgpt dot com or via the application to run in the Cloud shows up as “Could not load this ChatGPT conversation”.
Weirdly this does not always happens, but happens very frequently for me. Sometimes the same conversation will load, sometimes it shows this error (more often than not for me, it shows the error).
Reproduction
Steps or a minimal example prompt:
Literally any prompt run against ChatGPT Work in the Cloud reproduces the issue for me.
I do not know if it is related to the large number of custom SKILLs that I have added to my ChatGPT Cloud setup
Diagnostics
Exact error or screenshot:
When it happened (with timezone) and how often:
The intermittent part makes this a useful loader/state specimen rather than just “the conversation was deleted.”
You have the same Work conversation sometimes opening normally and sometimes returning “Could not load this ChatGPT conversation”, across Web/iOS/macOS.
I would preserve a small A/B before changing the account configuration:
exact affected conversation URL/thread reference;
where it was originally created;
whether the history/sidebar still lists it when loading fails;
timestamp + surface for one failed open;
timestamp + surface for one successful open of the same conversation;
whether a newly created Work conversation loads normally at the same time;
if available, the HTTP status/error for the failed conversation-load request, with auth/cookies removed.
I would not remove your custom Skills yet. Right now “many Skills are installed” is only a hypothesis. Changing that destroys the cleanest control unless you already have the failing state captured.
Also avoid deleting/recreating the affected Work thread just to recover access. If the same identifier alternates between success and failure, that is much more informative than a replacement conversation working.
The second Pro report here is useful independent corroboration, but it would be even stronger if that user can say whether the same conversation also intermittently loads rather than failing permanently.
ChatGPT keeps failing across my devices, even when I start a simple chat
I’m trying to find out whether other people are experiencing the same problem.
ChatGPT has been consistently failing for me on the website, the mobile app, and the desktop app. I can’t reliably open or use chats. Even trying to start a simple new chat fails, and I can’t upload files. I’ve tried accessing ChatGPT on different devices and internet connections, but the problem continues.
I reproduced the problem on the ChatGPT website in Chrome. The page showed “Something went wrong.” I checked the technical details, and ChatGPT’s server had returned an error while trying to process the conversation. The OpenAI Status page showed no active incident when I checked.
I’ve already captured the browser information from that failed attempt and can provide it privately to Support. I won’t post my conversation link or diagnostic file publicly because they may contain private information.
Anyone else having trouble opening or starting chats across the website and apps?
Did it resolve on its own, or did you find a fix?
Since today i also get a whole lot of “Could not load this ChatGPT conversation” when opening old topics/chats/work. Some open some not, the ones that open after a while they do not open, the ones that did not open then will start opening again, it seems random and there is no pattern.
But it’s very frustrating.
It happens on both PC(Firefox and Chrome) and mobile phone in ChatGPT App, both on ChatGPT Plus user, I cleared cache signed out etc but nothing changed.
In my case, I’m having the opposite issue: Work is the only feature I can’t use. I keep getting the error message, “Error in message stream,” whenever I try to use Work.
I’m a Plus user. Regular Chat still works, but GPT-5.6 has been behaving strangely for me, so if Work — where I can use GPT-6 — is unavailable, I can’t get my work done.
This suddenly started happening about 30 minutes ago.
+1 I started experiencing the same behaviour this morning. OpenAI’s status recorded no issues, but tjere are at least 10 affected users now from what I see. I found no fix, it seems as it is account-related and not device/OpenAI-end related.
I found a reproducible local Linux desktop failure with a similar loading symptom. This may be separate from the cloud/web/mobile issue in this thread: my affected Work sessions were marked “Runs on your computer”, and ordinary Chat still worked.
On Fedora 44, both 26.924.20706 and 26.924.22138 left old and new local Work sessions spinning. The desktop process was not collecting successfully exited child processes such as ps and git. An isolated trace showed native startup replacing libuv’s SIGCHLD handler.
What worked on my machine: after a verified history/profile backup, a private copy of 26.924.22138 was changed to wait for app.whenReady() before starting the early startup-requirements path. The installed RPM and history stores were left unchanged. The same old conversation then displayed its existing messages, a new projectless Work received a reply, and Voice listened/responded.
This is an experimental, build-specific workaround, not an official fix or a proven solution for cloud conversations. It changes startup ordering, so it still needs review by the app maintainers. It is not a browser-console command. Reverting means closing the private copy and launching the unchanged installed app; no history reset is needed.
If you are reporting the Linux variant, mentioning unreaped/zombie child processes and SIGCHLD handler replacement may help route it to the right investigation. I have not established that this explains the cross-device failures reported here.
Same here I’ve been having the exact same issue for the past day and a half..it’s ridiculous I can’t do anything without Chatgtp sending me the “Could not load this ChatGPT conversation” error…
I paid good money for pro and apparently it was a huge mistake.
Keep in mind I’m not the most techy guy, as I’m using ChatGPT Plus to mainly generate images,
When I generate an image, instead of showing the normal generation progress, it just stays on “Thinking” indefinitely. If I refresh the page, I often end up with an error instead of the generated image.
I can sometimes only generate one image every 5 mins, then problem is back. Both web and codex app say:
*“Could not load this ChatGPT conversation” *
Whereas mobile app says:
*“Too many requests” *
When going through my chats
The weirdest part is that it works perfectly fine when i switch to an alternate free account, is this problem account based?
I’ve also been having the same issue! 5.6 Sol is behaving weirdly, getting a range of different errors ranging from check your network to could not load conversation.
Not using Work, I am using the Chat mode. I have a Plus subscription. Is there an ongoing issue @openai? Or is this just throttling?
Also, when you send a message—it gets sent, and the button icon changes to the “stop” symbol—but the “Thinking…” text doesn’t appear. If you refresh the page, your message isn’t there.
