I completely agree. For simple or short conversations, the models are similar enough, even for the low parameter models. But as the conversation gets more compl

I completely agree. For simple or short conversations, the models are similar enough, even for the low parameter models. But as the conversation gets more complicated, the difference between the models becomes more obvious. The benchmarks used today only give a basic understanding of the language model's abilities and you have to use it yourself to see how capable it really is.