for function calling? I use gpt-4o and never had an issue with it. And llama 3.2 for conversation as it faster but bad with func calling. In my experience