Microsoft Phi-4

Phi-4 is Microsoft's 14B-parameter small language model that excels in complex reasoning, particularly in math, while optimizing size & computational efficiency. Available on Azure AI Foundry & Hugging Face, it sets a new benchmark for AI performance at scale.

Microsoft Phi-4 is a cutting-edge 14-billion-parameter small language model designed for advanced reasoning, particularly excelling in mathematics. Optimized for size and computational efficiency, it delivers high-performance AI solutions, making it ideal for scalable applications. Part of the Phi model family, Phi-4 is available on Azure AI Foundry and Hugging Face, offering cost-effective AI capabilities at the edge. The Phi series includes specialized models like Phi-4-multimodal for text, audio, and image tasks, Phi-4-mini for natural language prompting, and Phi-3.5 for multilingual support. These models collectively push the boundaries of generative AI, catering to diverse needs such as coding, language understanding, and reasoning, while setting new benchmarks in AI performance.