I manually selected GPT-5.5 thinking, but every response comes back as “used GPT-5.4 thinking” with no way to switch back. This has been going on for 8 days.
I exported my data and found this in the conversation JSON:
default_model_slug: gpt-5-5-thinking
resolved_model_slug: gpt-5-4-thinking (every single message)
Can anyone who understands the data export format confirm whether this indicates a backend routing mismatch or A/B test?
I’ve contacted support and they said it would be escalated to a specialist team, but it’s been 4 days with no follow-up.
I see that you have contacted support, but I was curious about this myself. I know that the way models are selected changes frequently, so one of the better places to check is the OpenAI ChatGPT FAQ pages.
This topic came up in my search, although it left me with even more questions.
Thanks for sharing the export details, that’s actually really helpful context. Since you’ve already got a ticket escalated, could you share your case ID ? I can take a look internally and see if there’s any update or additional signal on what’s happening here.
Thanks for looking into this, my case ID is 08365650.
Additionally I’ve seen other users report being occasionally routed to 5.4 mid-conversation, but they’re able to switch back to 5.5-Thinking by reselecting the model or starting a new chat. In my case, I cannot access 5.5-Thinking at all, reselecting the model and starting new conversations still routes to 5.4 every time.
Thanks for sharing the case ID and the additional context around the model behavior. I can see that your case has already been escalated to our specialized support team, and they’ll get back to you as soon as possible with any updates.
I appreciate the extra details you shared here as well.
Hi Smith, since your last reply I still haven’t received any further update from the support team, and the issue remains unresolved.
I’ve been completely unable to use GPT-5.5 thinking since April 29th. Every response across all conversations, devices, and new chats is generated by GPT-5.4 thinking, despite 5.5 thinking being selected. Your Help Center states that selecting Thinking means using GPT-5.5 Thinking, so I don’t understand why I am completely unable to access the model I selected.
At this point I need to understand what is happening with my account. Is this a known bug, an A/B test, or some form of account-level restriction? Could you please check and confirm?
Hi, I’d just like to add that I’m having the same problem, and I’ve seen other people on Reddit reporting it as well.
I have three observations. The first, and the strangest one, is that in the phone app (Android, in my case), the message shows as generated by 5.5 (Thinking-Extended), while in the web browser the very same message is shown as 5.4 Thinking.
The second point is less conclusive, but it still suggests something is off. When I ask the model what it is, based on its instructions, it consistently says the model I selected. I tested this on at least 20 messages, and it was correct every time, regardless of what the retry button said. When 5.5 is selected, it says 5.5 even though it shows 5.4. When I switch models within the same chat, it correctly says 5.2 or 5.4.
Another strange thing is that when the instant model uses thinking automatically, it correctly shows 5.5. However, when I use the instant model and then force a retry with thinking, it starts showing 5.4 instead.
I assumed this might just be a UI bug, especially because of what the mobile app shows, but I’d still feel better seeing a response from that “specialized support team.” Even if there’s a chance it’s only a UI issue, it still bothers me.
I’m a translator and use GPT to assist with my work, so I’m very sensitive to changes in tone and style. Everything was completely normal on the 28th. On the 29th, the responses suddenly shifted, and I’m confident they are coming from 5.4 Thinking, not 5.5. This isn’t just a feeling, my data export confirms it. The default_model_slug is set to gpt-5-5-thinking, but the resolved_model_slug on every single message shows gpt-5-4-thinking (screenshots in my original post).
As for support, my case was escalated to a specialist team over a week ago with no follow-up. After two follow-ups from me, they replied three days ago saying the internal team was reviewing, but there’s been no progress or clarification since. It’s been 11 days now and I still have no idea whether this is a bug, a test, or something else entirely.
This issue has now been ongoing for two weeks. More than a week ago, support told me the case had been escalated to the specialist team, but I still have not received any substantive update. A few days ago I was told it was still under review, and since then I have not received any further response, including after following up through the Help Center to ask whether the case is still being actively worked on.
I do not understand why the review is taking this long. From my side, I still have no visibility into whether there has been any actual progress. For the past two weeks, I have been completely unable to access GPT-5.5 Thinking, and every single message has been served by GPT-5.4 Thinking.
My subscription had already expired, and I resubscribed specifically to test whether this issue had been resolved, but the problem still persists. I consider this a serious issue, because I am paying for access to a model that I still cannot use.
Could someone please confirm whether my account has been incorrectly downgraded or misrouted, and whether this case is still under active investigation? My case ID is 08365650. I would really appreciate a clear update. Thank you.
Hi! I think they fixed this “bug.” I’ve finally been able to select 5.5 Thinking, and the retry button says 5.5! I hope your chat is also fixed. Also, OpenAI Support responded, so if for any reason you still have this bug, I’ll write to them and won’t close my ticket until they resolve your issue.
Hi! Thank you so much for letting me know. Mine also seems to be fixed now, and the retry button finally shows 5.5 Thinking on the web too.
I really appreciate you offering to keep your ticket open for me. That was so kind of you, and it honestly made me feel a lot less alone with this issue. I’m glad it seems to be resolved for both of us now. Thank you again!
