Try o3. My (very limited) experience with it is that it is refreshingly free of the normal LLM flattery, hedging, and overexplaining. It figures out your answer and gives it to you straight.