yeah its solved for me (after 10 hours) and thinking mode is working.
yeah both my ipad and iphone work fine but on desktop it is still saying either using 4 or 4 mini…
that was the entire point of this thread..?
I am still waiting for the fix, so far it is not working as expected…
The GPT on browser returns everything all based on 4o or mini 
Same, Chrome is abnormal, but it works normally after switching to Firefox.
I guess the reason may be the frequent replacement of network nodes.
I tried edge and it also works fine.
I believe it is the issue between chrome and backend.
I tried edge, chrome, and Firefox, all are not working
Having this same issue since yesterday! It does not matter if I switch between standard, extended, auto, they all just have instant output with no thinking whatsoever. If you ask the model which model we are using you get:
Which explains a lot, why it won’t think, why it skims over longer prompts and ignores most instructions, etc.
I also tried a customGPT that relies on the 5 thinking model, usually the ruleset replies with a very strict dataset for the user to provide, once provided it then goes on to think for anywhere between 2 - 5 minutes, and outputs a very specific result using set column names. But if you start the customGPT now, it just says rubbish like:
To help you effectively, it would depend on what you’re trying to do! Here’s what I might need in different situations:
General Questions:
- Just let me know what you need help with, and I’ll do my best to provide an answer.
Data Analysis:
- If you want to analyze or visualize data, feel free to share your dataset (in CSV, Excel, or another format), and let me know what kind of analysis or visualizations you’re looking for.
Creative Tasks (e.g., Writing, Brainstorming, etc.):
- Tell me what you’re aiming to create (an article, a story, an idea for a project), and any specific details or guidelines you’d like me to follow.
Research or Info Gathering:
- Let me know what topic you’re interested in and any specific aspects you’d like more information on.
Programming/Code:
- Share the code you’re working on, any errors you’re encountering, and describe what you’re trying to achieve.
If you’re not sure exactly what you need, you can just ask, and I’ll help you figure it out!
But the customGPT has a predefined set of instructions, so it should not respond this, especially when you consider this is the response from clicking one of it helper buttons. If I try to continue after that, it just blurbs loads more rubbish and the customGPT doesn’t even work at all.
According to the status “it should be operational”. But, I disagree:
Things I have tried:
- Disabled “Developer Mode”
- Incognito
- Disabled Caching in the network tab
- Deleted all cookies, site data, and reset the site permissions
- Deleted all browser history
- Tried a new chrome profile
- Flushed DNS
- Reset router
- Disabled VPN
- Disabled any AD Blockers & Anti-virus, etc.
- Tried on a mobile device (same result)
- Tried on a laptop (same result)
- Even tried a virtual machine running Linux
The same outcome, no matter what is used.
Kind regards,
Gunner
I encountered the same issue. Yesterday, I registered for ChatGPT and subscribed to Plus. When using it, I manually selected the ChatGPT 5 Thinking model, but it didn’t show any thinking process when I asked questions. When I asked about the version, it replied “ChatGPT 4 Mini”. Later, I tried using the ChatGPT app on my Apple tablet and found it could work normally with ChatGPT 5 Thinking. However, after a few hours, the tablet stopped functioning properly again, and no questions showed any thinking process. During this period, I tried logging in through different browsers and repeatedly deleted cookies, but none of these solutions worked to resolve the ChatGPT 5 Thinking model issue.
I encountered the same issue.
I managed to fix it by asking for certain info he would need internet browsing to find, then I asked him “Source?”, and then miraculously he reverted back to thinking. Wtf
Funny Fact If I enter through this link from OpenAI to Chatgpt: Apresentamos o GPT-5 | OpenAI the first Prompt I try It uses GPT-5 mode, after that if goes straight to mini. Can we all get a fix?
I seem to be suffering from this issue as of yesterday! Seems like all my queries get routed to a mini, non-thinking model. Smells a lot of gpt-4o or similar models. Weirdly the models will sign off as the correct model, but the capability change is very obvious (and very frustrating)
Same issue here. 28.10.25. I tried with edge an it worked with 5 thinking model.
same issue here. 28.10.29
Same problem, and I see that this chat has been around since August. It seems that no developer is going to help us.
This is getting ridiculous. Not even a single answer by the devs.
This has been going on for months, and yes I have the same issue as everyone else here.
Is this the buggy mess that is GPT 5?
Though I am 99% sure this is 100% intentional. They root the base model to cheaper ones to safe money, even though we paid for the model. A Scam.
Seriously WTF is happening this is ridiculous, I’ve held out far, unsubscribing from Pro right now.