Hi!
I asked ChatGPT a question in Portuguese, and the answers came with the internal chain-of-thought in English with faded letters, followed by the answer in Portuguese in clearer letters.
That shouldn’t be shown, right? Could it have been a temporary bug?
3 Likes
Same for me, but just English/English on 4o. Also the responses sounds like o1 - ad-nauseum over-thought, over-wrought and even wrong. I asked it to look up and comment on DeepSeek inference and it missed the news. Wtf.
Same here, always reasoning.
It’s exactly the same for me. It makes the responses much longer and even buggy. I have to log out on my smartphone, clear the cache and memory to temporarily avoid this bug. It’s really driving me craz
1 Like
This was even happening with my account. Did the steps you mentioned and thankfully now it was back to normal. Thanks for telling about this, hopefully doesn’t happen again!! 
EDIT: Just after 5 minutes, this is repeating again. Seems like even after choosing 4o as the model, it is using o1 Model. Hopefully, OpenAI works on this bug. Because some tasks do not require reasoning and it just takes a lot of time and unnecessary analyzing, even when I greet the model.
The only solution I was able to get for now was to use ChatGPT from the Web. Whenever, I am using ChatGPT from my Laptop (because I use it from the web), it is not reasoning always. For now, this works or if you want it to temporarily work, then removing cache also works but for a chat or two.
1 Like
Yes, if you’re on a smartphone, clearing the app’s cache and storage temporarily fixes the issue, but it reappears shortly afterward, which is very annoying. The alternatives are using it on a web browser or a computer. We’re not the only ones experiencing this bug, so let’s remain patient and endure this inconvenience for now (´Д`)
1 Like
True! This is isn’t happening with the free accounts yet, but I guess the bug is due to the Recent Update of ChatGPT (Beta). I am sure OpenAI will deal with this, and hopefully soon 
Solved on Android with today’s update. I was worried because that behavior made the difference between high daily value and actual worthlessness.
Can’t confirm that, ‘cause I’m still havin’ the same issue even after installin’ the latest version. I even uninstalled and reinstalled the app, but the bug’s still not fixed on my end.
The new version I have has both Search and Reason buttons. The previous had only Search.
Before the update, in one instance I got a o1 warning of 25 messages remaining, even though I was supposedly running 4o. I think that was the bug - not selecting models correctly.