and is extremely slow - almost useless for me comparing to previous versions. I would still prefer to choose explicitly GPT-4 model that doesn't browse and don't do image recognition.
Now too often it goes to searching the web even if sometimes I don't need up to date information e.g. about GDP of some countries I just only need some ballpark stats.
> I would still prefer to choose explicitly GPT-4 model that doesn't browse and don't do image recognition.
I have disabled it in settings. Which only decreased usage of it. Which is ridiculous, especially that you can avoid using e.g. "browsing" by explicitly stating it in the input.
Free cgpt users and low tier API users are downgraded opaquely to lower models when demand is high.
It's irritating because I have over 4000 API chats in a DB that I like to analyze, but I cannot now know for certain which model was actually being used.
Ohh so that explains why I saw `?model=text-davinci-002` appended to my ChatGPT URL the other day. I remember it was pretty slow to load then, too, so that's probably why they rerouted me
Now too often it goes to searching the web even if sometimes I don't need up to date information e.g. about GDP of some countries I just only need some ballpark stats.