intel/inference-xeon-microsoft-phi-4-reasoning repository overview

⁠Intel® Inference Microservices

This is an Intel® Inference Microservices model image. It serves one large language model on Intel® silicon through an OpenAI-compatible API. At startup the container applies the settings Intel® tuned for this model.

⁠How to run it

Copy the complete docker run from this model's listing in the Intel® Software Catalog⁠. Open the model, then use the Deployment tab.

That command is the supported way to start Intel® Inference Microservices. It already has the image name, flags, cache mount, and — when the model is gated — HF_TOKEN. Do not assemble a command from this Docker Hub page.

⁠After it is running

Point any OpenAI-compatible client at http://localhost:8000/v1 (OpenAI SDK, LangChain, LlamaIndex, LiteLLM, Haystack, or curl). Wait until GET /health returns 200.

⁠Supported silicon

Runs on Intel® silicon. Which hosts this image supports: Supported Intel® platforms in the Intel® Inference Microservices documentation.