Phi-3-small

Phi-3 Small is a dense ~7B-parameter model in Microsoft’s Phi-3 family, designed to deliver high quality at low latency and cost. It’s text-in/text-out, instruction-tuned, and strong on practical reasoning, multilingual tasks, and code generation. The model supports function/tool calling, JSON-formatted responses, and integrates cleanly with RAG and agent frameworks. Open weights and a permissive license make it easy to run on Azure AI or locally; 8-/4-bit quantization and LoRA fine-tuning help it fit tight memory budgets. Typical uses include chat assistants, document QA/summarization, coding helpers, and on-device prototyping where efficiency and reliability matter.