Ultralytics, the developer of the YOLO family of object detection models, has announced a collaboration with Intel Corporation (NASDAQ: INTC) to optimize its latest YOLO26 computer vision models for Intel processors, aiming to simplify deployment of real-time vision AI across edge computing applications.
The collaboration combines Ultralytics’ production-ready YOLO models with Intel’s OpenVINO Toolkit, enabling developers to deploy computer vision applications on Intel-powered CPUs, GPUs and neural processing units (NPUs) without requiring discrete graphics processors.
The companies said the initiative targets industries where vision AI is increasingly deployed on existing CPU-based infrastructure, including industrial PCs, laptops and edge devices.
By optimizing the models for Intel hardware, the collaboration is intended to reduce deployment costs while improving inference performance. According to the companies, supported CPU and GPU configurations can achieve inference latency of less than five milliseconds across YOLO tasks.
The announcement reflects a broader shift toward edge AI, where machine learning models are processed closer to where data is generated rather than relying solely on cloud infrastructure. This approach can reduce latency, improve data privacy and lower operating costs for industrial and commercial applications.
Glenn Jocher, Founder and Chief Executive Officer of Ultralytics, said most enterprise AI models are trained in data centers, but production computer vision applications typically run on edge devices powered by Intel processors.
He said the collaboration enables developers to deploy production-ready YOLO26 models on existing Intel hardware, eliminating the need for dedicated GPUs in many applications.
The companies also said the partnership simplifies AI development workflows. Developers can train models using the Ultralytics platform and export them directly to OpenVINO for deployment through the same Python package or command-line interface, reducing the complexity of moving AI models from development to production.
The optimized models are expected to support a broad range of applications across multiple industries.
In manufacturing, the technology can be used for quality inspection, defect detection and production monitoring. Logistics providers can deploy the models for parcel detection, asset tracking and automated counting, while retailers can use them for shelf monitoring, inventory management and product recognition.
Other target applications include surveillance and compliance monitoring in the security sector, medical imaging workflows in healthcare, real-time perception systems for robotics, and distributed computer vision infrastructure supporting smart city deployments.
The collaboration also supports broader edge AI use cases by leveraging integrated AI acceleration already available in Intel system-on-chip platforms.
Matthew Formica, Senior Director and Global Head of Edge Technical Marketing at Intel, said the collaboration expands Intel’s AI PC and physical AI ecosystem by making leading open-source vision models available on processors with built-in AI acceleration.
He said optimizing YOLO26 for the latest Intel Core Ultra processors and future platforms will help developers deploy AI applications more efficiently across real-world environments.
