On Codeforces competitive programming, @OpenAI o3-mini achieves progressively higher Elo scores with increased reasoning effort, all outperforming o1-mini. With medium reasoning effort, it matches o1’s performance.

→ View original post on X — @romainhuet