As the title says, we get an option to choose which GPT model (4 or 3.5) we want to use for the rest of a chat session either in the settings or a prompt after starting a new chat session.
Instead of defaulting to GPT-4 and the chat later switching to GPT 3.5 after we exhaust our GPT-4 time.
It makes the chat more coherent and less messy when we have the option to stick to one model for the rest of the chat.
(Free GPT-4 users can continue their chat session after each cool down period from where they left off).