Hi! I encountered a problem - GPT 5 Thinking mode does not work as it should on the desktop and mobile app. The model responds instantly, without a delay for thinking, although everything was fine before. I use the same account, and the problem is the same on all devices. Also, none of the models (including GPT-o3) work properly — they only give instant responses, with no thinking delay at all. I tried reinstalling the browser, clearing the cache, running in incognito, checked on different browsers and reinstalling the application, but nothing helped. Maybe someone encountered such a problem and knows how to fix it?
When I asked what the model was, he replied, “It’s a GPT-4 model, with a smaller version (mini) for faster response, trained by OpenAI”, although at this moment gpt 5 thinking was chosen.
And yes, this is GPT Plus
Та же самая проблема, gpt 5 пишет что gpt 5 на данный момент не существует, а он является gpt 4. Так же он абсолютно не следует инструкциям в проектах и созданных gpt
From the phone: If you select gpt 5 thinking, it gives 4 mini. If you select auto, it works 4 mini. If you ask auto to think, it thinks like 5 thinking. If you select gpt 5 thinking and ask to think, it gives mini.
From desktop - doesn’t matter, always mini. GPT 5 Thinking on the phone call it “routing/feature-flag regression”
i have the same problem today (Aug 30).
all the models answers immidiately without thinking and they are too weak now!!!
wee need clear answer from open ai
same problem here. They answer immediately whatever model I pick
Same; I am a Pro user. It cannot use the the thinking function since 6 hours ago and it is no use to email for technological support.
Tech support requested a video recording of the work and responses of the gpt, I sent it. There is no solution yet
Yeah, I can’t use ChatGPT 5 without being able to use the Thinking version.
im having the same problem and its driving me mad. i cant find a solution so ive had to halt work on my project. i think it is going to be a case of sitting and waiting until it starts to work properly again. Im guessing that it is a technical issue as even though im on plus subscription there is no way that i have even come close to the 3000 message weekly limit that they say applies so it cant be that. i have tried rebooting my system etc it didnt make a difference. if you find a cure for it then can can you please post it on here thanks 
I am also experiencing this issue. First, it appeared on Firefox when canvas stopped working → switched to Chromium-based → worked → stopped working again → tried it on the phone → worked → stopped working.
I used the prompt below, then regenerated it with all the models available when you enable legacy models. All of them signed off as 4o… lol.
It is not simply hallucination:
- GPT-5 can normally correctly identify itself.
- There was no thinking process present.
- Instant answers.
- No (“Thinking for a better answer”) label thingy.
- “All” models had the same generation speed, the same speed that I am familiar with in 4o.
- Overall style is clearly the 4o model.
- This list is stupid; it’s pretty obvious.
- Stop reading.
- Fix the issue, please. Thank you.
I have this issue too.
I don’t know what happened. And it just wasted my “Agent mode’s.
I strongly believe “openai” should consider this global issue and recharge used limited modes which were used during this global issue!
Same issue here. Just upgraded to the team plan and was eager to try out the gpt 5 pro model, and was confused why it wasn’t thinking. Curious, I went to o3, gpt 5 thinking mini and thinking, and none of it worked. Kinda sad considering we don’t get that many prompts for the heavy models and that I probably wasted a ton of prompts.
I have the same issue, 5 only works on my phone. If 5 is down on desktop browsers, update the label. Kinda typical OpenAI… gaslight the users into thinking they are imagining stuff
hopefully the backend detects the requests as what they actually are, meaning they wont subtract from your usage.
I am also facing this issue and am very worried as I am no longer able to work. Any Update how it can be resolved? Where to report this issue? Its really frustrating.
Seems like they already fix it!
no, its very inconsistent is all. sometimes auto works, sometimes instant works, sometimes thinking works.. lol. mostly not.
It works one time then again switches to 4.0 
seems to be fixed now, for me at-least.
how is it for everyone else?
