Introducing NVIDIA NIM Microservices and AI Blueprints to Usher in a New Era of Local AI.
NVIDIA RTX AI PCs
In this first edition of the RTX AI Garage series, additional CES announcements are summarized, along with new NIM microservices and AI blueprints that enable generative AI on NVIDIA RTX AI PCs and workstations.
Industry-defining technological innovations have been created in garages many times. For developers and fans interested in learning more about NVIDIA NIM microservices and AI blueprints, as well as how to create AI agents, digital humans, productivity apps, and more on AI PCs, the RTX AI Garage series begins this week. The RTX AI Garage welcomes you.
The developments made earlier this week at CES are highlighted in this first chapter, including new AI foundation models that are available on NVIDIA RTX AI PCs and that elevate digital humans, content creation, productivity, and development.
The latest GeForce RTX-50 Series GPUs power these models, which are available as NVIDIA NIM microservices. The RTX 50 Series GPUs, which are based on the NVIDIA Blackwell architecture, have 32GB of VRAM, FP4 computing, and can do up to 3,352 trillion AI operations per second. This doubles the performance of AI inference and makes it possible for generative AI to operate locally with less memory usage.
Additionally, NVIDIA unveiled NVIDIA AI Blueprints, which are preset, ready-to-use workflows for applications such as content creation and digital people that are based on NIM microservices.
With the help of NIM microservices and AI blueprints, developers and enthusiasts can create, refine, and release AI-powered PC experiences more quickly than ever before. As a result, PC users now have access to a new generation of intriguing and useful features.
Accelerate AI Using NVIDIA NIM
Introducing AI developments to PCs is fraught with two major obstacles. First, new models are being created every day on sites like Hugging Face, which currently has over a million models, demonstrating the rapid speed of AI development. Consequently, innovations are soon rendered obsolete.
Second, it takes a lot of work and resources to modify these models for PC use. It takes a lot of engineering work to connect them to programs, integrate them with AI software, and optimize them for PC hardware.
Through the provision of preconfigured, cutting-edge AI models tailored for PCs, NVIDIA NIM assists in addressing these issues. These NIM microservices use RTX GPUs and NVIDIA AI software to boost performance, span model domains, are single-click installable, and include application programming interfaces (APIs) for simple integration.
Large language models (LLMs), vision language models, picture generation, voice, retrieval-augmented generation (RAG), PDF extraction, and computer vision are among the application cases that NVIDIA’s pipeline of NIM microservices for RTX AI PCs supports, as the company stated at CES.
On a variety of agentic tasks, the new Llama Nemotron family of open models offers excellent accuracy. For NVIDIA RTX AI PCs and workstations, the Llama Nemotron Nano model will be available as a NIM microservice. It is highly proficient in agentic AI tasks including as math, coding, conversation, function calling, and command following.
These microservices will soon be available for developers to download and use with Windows Subsystem for Linux (WSL) on Windows 11 PCs.
NVIDIA showcased Project R2X, a vision-enabled PC avatar that can read and summaries texts, help with desktop programs and video conference calls, and more, to show how hobbyists and developers can leverage NIM to create AI agents and assistants.
AI enthusiasts can avoid the hassles of model curation, optimization, and backend integration by utilising NIM microservices, allowing them to concentrate on developing and experimenting with state-of-the-art AI models.
What is an API?
An application’s interface with a software library is called an API. A collection of “calls” that an application can make to a library and the expected responses are specified by an API. The extensive setup and configuration needed for traditional AI APIs limits innovation and makes AI capabilities more difficult to use.
NIM microservices provide applications with simple, user-friendly APIs to which they can submit queries and receive a response. Additionally, they are made with the input and output media for various model kinds in mind. For instance, voice recognizers turn speech into text, picture generators turn text into images, LLMs take text as input and output text, and so forth.
Leading AI development and agent frameworks including LM Studio, AnythingLLM, ComfyUI, Flowise AI, LangChain, Langflow, and AI Toolkit for VSCode can all be easily integrated with the microservices.
By making these APIs available on RTX, NVIDIA NIM will hasten the development of AI on PCs. With the next iteration of the NVIDIA ChatRTX tech demo, enthusiasts should be able to test out a variety of NIM microservices.
Get Started With the New AI APIs for PCs
Beginning next month, NVIDIA NIM microservices and AI blueprints should be accessible, with first hardware support for professional GPUs like as the NVIDIA RTX 6000 and 5000, GeForce RTX 50 Series, and GeForce RTX 4090 and 4080. In the future, more GPUs will be supported.
It is anticipated that local system builders Corsair, Falcon Northwest, LDLC, Maingear, Mifcon, Origin PC, PCS, and Scan, as well as Acer, ASUS, Dell, GIGABYTE, HP, Lenovo, MSI, Razer, and Samsung, would offer NIM-ready NVIDIA RTX AI PCs .
GeForce RTX 50 Series laptops and GPUs power transformative AI experiences, give game-changing performance, and allow creative finish workflows faster than ever before.
