Microsoft Phi-3 GenAI Models

Accelerated by Intel AI Platforms

Intel Partners with Microsoft to Boost Phi-3 Model Performance

Intel has partnered with Microsoft to optimize several Phi-3 models for its edge solutions, AI PCs, and data center systems. This collaboration ensures that Microsoft’s Phi-3 family of open models runs smoothly across Intel’s hardware ecosystem.

What’s New in the Phi-3 Integration?

Intel has streamlined and validated its AI product lineup to support the Microsoft Phi-3 family. These models are open-source, compact, and easily customizable. They run efficiently on less powerful hardware, enabling developers to build applications that can operate locally on edge devices and client machines.

Intel’s AI Hardware Supports Phi-3 Models

Intel’s supported products include:

• Intel® Gaudi® AI accelerators and Intel® Xeon® processors for data centers

• Intel® Core™ Ultra processors and Intel® Arc™ graphics for client devices

Pallavi Mahajan, General Manager of Data Center and AI Software at Intel, said:
“We provide customers and developers with powerful AI solutions using the latest AI models and software. Our collaboration with Microsoft is key to bringing AI everywhere. We’re proud to ensure Intel hardware actively supports several new Phi-3 models across data centers, edge, and client devices.”

Why This Partnership Matters

Intel consistently invests in the AI software ecosystem to expand AI adoption. Working with Microsoft, Intel enabled Phi-3 model support on launch day for its CPUs, GPUs, and Gaudi accelerators. They also co-designed DeepSpeed, a deep learning optimization toolkit, and enhanced Hugging Face’s automatic tensor parallelism for Phi-3 and other models.

Advantages of Phi-3 Models on Intel Platforms

Due to their small size, Phi-3 models are ideal for on-device inference. This allows lightweight AI model development on PCs and edge devices, including fine-tuning and customization. Intel accelerates this development through software frameworks such as PyTorch, Intel® Extension for PyTorch for local R&D, and the OpenVINO™ Toolkit for deployment and inference.

Looking Ahead: Supporting Business AI Needs

Intel remains committed to supporting Phi-3 and other advanced language models. The company focuses on software optimization to meet the generative AI demands of business clients, ensuring high performance and flexibility across Intel hardware platforms.

Intel acceleration of Microsoft Phi-3 AI models