NVIDIA's new foundation models are designed to enhance the capabilities of RTX AI PCs by supporting applications in digital humans, content creation, and productivity. These models leverage NVIDIA NIM microservices, which simplify the deployment of the latest generative AI models.

During CES, NVIDIA unveiled that these models will operate locally on its RTX AI PCs, featuring the newly announced GeForce RTX 50 Series GPUs. The GPUs boast impressive specifications, including up to 3,352 trillion operations per second and 32GB of VRAM, thus allowing generative AI models to run efficiently with a reduced memory footprint.

The history of innovation on NVIDIA's platform is significant: the GeForce GTX 580 was pivotal in the development of deep learning back in 2012, and recent estimates show that over 30% of AI research publications mentioned the GeForce RTX family for their work.

NVIDIA emphasizes accessibility, enabling developers and enthusiasts with low-code tools such as AnythingLLM and ComfyUI to utilize AI models through user-friendly interfaces. The introduction of AI Blueprints will provide preconfigured workflows for various applications, making it easier to integrate these features into PCs.

“AI is advancing at light speed,” said Jensen Huang, NVIDIA's founder and CEO. “NIM microservices and AI Blueprints give PC developers and enthusiasts the building blocks to explore the magic of AI.”

NVIDIA also announced a partnership with several model developers, leading to a pipeline of NIM microservices that encompass a broad range of use cases—from language and speech models to advanced computer vision.

A significant upcoming offering is the Llama Nemotron family of open models, which excel at various AI tasks like instruction following and coding. NVIDIA's microservices are optimized for deployment across both RTX PCs and cloud services.

Developers will find it easy to download and set up these microservices on Windows 11 PCs, supported by Microsoft’s Windows Subsystem for Linux. Notably, these microservices will be compatible with popular AI development frameworks, allowing for a seamless integration between applications and AI models.

NVIDIA aims to cater to both developers and enthusiasts with a range of experiences, including a tech demo for its upcoming ChatRTX platform. The potential applications of the NIM microservices are extensive, spanning graphic representation and automation within various workflows.

NVIDIA is set to roll out these innovations in February, with initial hardware support for the latest GeForce RTX 50 Series GPUs, among others. Major PC manufacturers are preparing to launch NIM-enabled RTX AI PCs to meet growing demands across sectors.

This initiative represents a significant step forward in making AI capabilities more accessible to a broader audience while driving innovation in personal computing.