Compare o4-mini (OpenAI) and Claude 3.7 Sonnet (Anthropic) extended thinking
Note: Reasoning models are slower but significantly more accurate on complex problems.