Is llama 3.3 enough? I think Gemini better