That's possible, my counter point would be that if that was the case Anthropic would have built a smaller reasoning model instead of doing a "full" Claude. Instead, they built something which seems to be flexible across different types of responses.

Only time will tell.