Comment by dvfjsdhgfv

The "open source" LLMs are already good enough for simple tasks GPT-3.5 was used for. I see no reason why they can't catch up with GPT-4 one day.