NVIDIA at the CES 2025 announced the launch of foundation models designed to run locally on NVIDIA RTX AI PCs, unlocking new possibilities for digital humans, content creation, productivity, and development. This move signifies a major shift, bringing the power of generative AI directly to the hands of users and solidifying NVIDIA’s commitment to democratizing AI.
GeForce: A Legacy of AI Innovation
The GeForce platform has long been a cornerstone for AI developers. The groundbreaking AlexNet, the first GPU-accelerated deep learning network, was trained on the GeForce GTX 580 in 2012. Last year alone, over 30% of published AI research papers cited the use of GeForce RTX, highlighting its enduring impact on the field.
From Enthusiasts to Developers: A New Era of AI Accessibility
With the advent of generative AI and RTX AI PCs, NVIDIA is opening the doors to AI development for everyone. New low-code and no-code tools like AnythingLLM, ComfyUI, Langflow, and LM Studio empower enthusiasts to leverage complex AI models through intuitive graphical user interfaces.
Further simplifying the process, NVIDIA is introducing NIM microservices. These will connect seamlessly with popular GUIs, providing effortless access and deployment of the latest generative AI models. NVIDIA AI Blueprints, built on NIM microservices, offer preconfigured workflows for various applications, including digital humans and content creation.
Expanding the RTX AI Ecosystem
To meet the burgeoning demand, every major PC manufacturer and system builder is launching NIM-ready RTX AI PCs. These microservices will also be integrated into NVIDIA Project DIGITS, a personal AI supercomputer powered by the new NVIDIA GB10 Grace Blackwell Superchip. Project DIGITS, featuring a petaflop of AI computing performance, will empower researchers, data scientists, and students to prototype, fine-tune, and run large AI models.
Project DIGITS is priced at starting at $3,000 (Rs. 2,57,060 approx.) and will be available in May from NVIDIA and top partners.
Making AI NIMble: The Power of Foundation Models
Foundation models, the bedrock of generative AI, are neural networks trained on massive datasets. NVIDIA is set to release a comprehensive pipeline of NIM microservices for RTX AI PCs, featuring contributions from leading model developers like Black Forest Labs, Meta, Mistral, and Stability AI. This collection will encompass a wide range of use cases, including:
- Large Language Models (LLMs)
- Vision Language Models
- Image Generation
- Speech
- Embedding Models for Retrieval-Augmented Generation (RAG)
- PDF Extraction
- Computer Vision
NVIDIA also announced the Llama Nemotron family of open models, known for their high accuracy on diverse agentic tasks. The Llama Nemotron Nano model, available as a NIM microservice for RTX AI PCs and workstations, will excel in tasks like instruction following, function calling, chat, coding, and math.
NIM microservices are optimized for deployment across NVIDIA GPUs, ensuring seamless performance whether running on RTX PCs, workstations, or in the cloud.
Streamlined Development and Deployment
Developers and enthusiasts can quickly download, set up, and run these NIM microservices on Windows 11 PCs with Windows Subsystem for Linux (WSL). They will be compatible with leading AI development and agent frameworks, including AI Toolkit for VSCode, AnythingLLM, ComfyUI, CrewAI, Flowise AI, LangChain, Langflow, and LM Studio. This compatibility allows developers to connect applications and workflows to AI models running NIM microservices through industry-standard endpoints, enabling a unified interface across various platforms.
Enthusiasts will also be able to explore a range of NIM microservices through an upcoming release of the NVIDIA ChatRTX tech demo.
Project R2X: Bringing AI Agents to Life
To showcase the potential of NIM for building AI agents and assistants, NVIDIA previewed Project R2X, a vision-enabled PC avatar. This avatar can:
- Provide instant access to information
- Assist with desktop applications and video conference calls
- Read and summarize documentsR2X is rendered using NVIDIA RTX Neural Faces, a new generative AI algorithm, and animated by the diffusion-based NVIDIA Audio2Face™-3D model, which enhances lip and tongue movement. It can connect to cloud AI services like OpenAI’s GPT4o and xAI’s Grok, as well as NIM microservices and AI Blueprints, through frameworks like CrewAI, Flowise AI, and Langflow. Users can sign up to follow updates on Project R2X.
AI Blueprints: Ready-to-Use AI Workflows for PCs
NIM microservices will also be accessible through AI Blueprints, ready-to-use AI workflows designed for local execution on RTX PCs. These blueprints enable developers to create diverse applications, such as:
- PDF to Podcast: This blueprint extracts text, images, and tables from a PDF to create an editable podcast script. It can generate a full audio recording from the script using various voices or even a user’s voice sample. Users can also engage in real-time conversations with the AI podcast host to delve deeper into specific topics.
- 3D-Guided Generative AI: This blueprint offers artists finer control over image generation. By using simple 3D objects arranged in a 3D renderer like Blender, creators can guide the AI image generation process, achieving precise control over composition.
Availability
NVIDIA NIM microservices and AI Blueprints will be available starting in February. NIM-ready RTX AI PCs will be available from leading manufacturers, including Acer, ASUS, Dell, GIGABYTE, HP, Lenovo, MSI, Razer, and Samsung, as well as various system builders.
This announcement marks a significant leap forward in bringing the power of AI to a wider audience. By enabling local execution of foundation models on RTX AI PCs, NVIDIA is fostering a new era of AI-powered creativity, productivity, and development, limited only by the imagination of users and developers.