max/xhigh reasoning levels

Is more reasoning necessarily better?

A Reddit user questions whether using maximum or extra-high reasoning settings in AI models consistently improves output quality, noting increased token consumption and repetitive reasoning without clear evidence of superior results.