Interesting to see reasoning models lacking behind their non-reasoning counterpart