A reasoning-focused LLM from DeepSeek optimized for complex logical inference and step-by-step problem solving.