For close to 2 weeks now, o1-preview and o1-mini have both stopped “Thinking…”.
We have multiple paid accounts across our company (not Teams, individual accounts) and it’s happening with all of them.
When we enter a prompt/message, we get an immediate response, just like with GPT-4o and GPT-4. And the responses are clearly very inferior to what we used to get with o1-preview.
Even if we force /reason it still doesn’t work.
And no, before anyone says, we haven’t hit our message cap. We didn’t use o1-preview for a full week across all accounts, just to ascertain this. And still, after a full week, we still get instant GPT-4o-like responses from “o1-preview” with not even 1 second of thinking.
It’s incredibly frustrating since we’re paying for ChatGPT almost entirely for the privilege of using o1-preview.
Yes, I encountered it, but I have no idea what to do. Can I solve it by changing my account and paying? Or are there any other companies’ models that can compete with o1preview? Now o1preview’s answer is like GPT3.5
I just tested your prompt, 01-preview completed it in 16 seconds.
FYI - I know the share is not found, trying to figure out why and will update if possible.
I do sometimes get incomplete results but not so much so that it causes me problems. Often just retiring the prompt, or waiting about 5 minutes and trying again or such often works. However I will note that I do not use the o1 models just because they are there. Often I will use GPT 4o first and if that does not work give the o1 models a shot. There are a few things that I will start with the o1 models first such as asking it to create a self contained HTML page with HTML,JavaScript, CSS and JSON. These questions often with other models require a few more fixes before the code works as needed while the o1 seem to get it correct most of the time with the first version. Once the first version is working then I move back to GPT 4o.
My O1 Preview and Mini can’t function properly, whether using the official GPT software or various browsers — the context length is reduced from 40 to 40 Mini. As a result, I can only use the mobile version, which works fine."
I’ve encountered the same issue. May I ask if your problem has been resolved? I asked ChatGPT about your model version, and no matter which model I select, it always replies with GPT-4 Turbo.
Same here. My o1-preview hasn’t been thinking for over half a month. I even thought OpenAI had hidden the thinking process.
Damn, the experience for paying users is almost no different from free accounts now
I’ve been dealing with this issue for 2 days now. Yesterday, both o1-preview and o1-mini skipped the thinking phase, but today, o1-preview started ‘thinking’ again. This time, it’s thinking fast, responding fast, and super verbose, just like o1-mini. Based on the limit notification message, I’m pretty sure it’s o1-mini. So, I think 4o and o1-mini get ‘routed’ to 4o-mini, as I did some limit tests and never reached the limit for either of them. While o1-preview get ‘routed’ to o1-mini.
The way the thinking is presented is different. Before it use to show you the thinking steps in the conversation. Now thinking is a link that if you click on it will show the thought process in a pop up window.
