[energy123]

This was mentioned in OpenAI's report. People rated o1 as the same or worse than GPT-4o if the prompt didn't require reasoning, like on personal writing tasks.