i relize that 4.0 tends to be less correct then 3.5 why is that
Hi!
Can you share some examples of your prompts and maybe your custom instructions? If you actually did find some model degradation it’s going to be interesting.
Otherwise there’s the option of user error 
sure i asked it to solve a complex fraction which I knew the answer was 4.5 and It told me the answer was 4.9 and I put the complex fraction into the fraction.net calculator and got 4.5 but I asked 3.5 gpt and got 4.5 as the answer also vb can you read my message to the mods team? @vb
There are many topics in the community as to why the models are not good at maths without proper tool support.
I can only read messages that were send to me. Fortunately.
ok @vb so i will direct the message to you
edit I directed it towards you
This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.