By skyforbes Does anyone know what model chat GPT actually uses when you are not using the “thinking” model? The reasoning models of 5 and 5.1 are pretty incredible, but when I switch to “instant” in chatGPT, it goes full AI slop mode. Hallucinations, emojis everywhere.
I’ve been using 4o when I want quick answers, because it seems to be more accurate than the “instant” 5 models and you can use custom instructions to make it less annoying/sycophantic. I can’t find any benchmarks to prove that 4o is a better model than whatever model chat GPT is using for the instant answers. All of them just compare 4o to the full version of 5 which is obviously going to score better.