Phi-4 Reasoning introduces a new class of small, open-weight models, available in 3.8B and 14B parameters, engineered to deliver exceptionally powerful reasoning capabilities across complex domains such as mathematics, scientific inquiry, and code generation. These models demonstrate performance that rivals much larger Large Language Models (LLMs), offering efficiency without compromising on advanced problem-solving. Phi-4 is readily accessible, allowing developers and researchers to leverage its capabilities on both Azure AI Foundry and Hugging Face, democratizing access to cutting-edge AI for specialized tasks.