Claude 3.5 is worse than Gemini 1.5 pro

I thought claude was meant to be better than gemini since it generally is in the llm benchmarks but claudes code explanation is really bad and doesn’t sound human at all meanwhile gemini 1.5 pro is and provides very in depth explanation