At the Consumer Electronics Show (CES) 2025, Nvidia unveiled its ambitious Nemotron Model Families, which are poised to revolutionise the field of agentic AI. This ground-breaking announcement was made during the keynote address delivered by Nvidia’s CEO Jensen Huang. The company introduced open Llama Nemotron large language models (LLMs) and Cosmos Nemotron vision language models (VLMs), designed to enhance AI agents operating on any accelerated system.
Nvidia's focus on agentic AI underscores a new era in artificial intelligence where specialised teams of agents can aid individuals in tackling complex challenges and automating repetitive tasks. Huang emphasised the potential of AI agents, stating, “AI agents is the next robotic industry and likely to be a multibillion-dollar opportunity.” This sentiment reflects a broader industry shift towards leveraging AI to amplify productivity across various sectors.
The Llama Nemotron family aims to provide enterprises with a robust foundation for implementing agentic AI solutions. These models, constructed using the Llama framework, will enable developers to create and integrate AI agents into numerous applications, encompassing areas such as customer support, fraud detection, and supply chain management.
To function effectively, many AI agents will necessitate both linguistic capabilities and the ability to perceive their surroundings. The newly introduced Cosmos Nemotron VLMs and Nvidia NIM microservices for video analysis will allow developers to craft agents capable of interpreting and acting on visual data sourced from various environments, including autonomous vehicles, hospitals, and retail spaces.
The Llama Nemotron models, built on Nvidia’s acclaimed open-source model collection Llama—which boasts over 650 million downloads—are optimised for compute efficiency and accuracy. According to Nvidia, these models have undergone new training techniques and pruning processes, resulting in significant enhancements in instruction adherence, conversational abilities, and mathematical functionalities.
Ahmad Al-Dahel, vice president and head of GenAI at Meta, commented on the evolution of AI capabilities by stating, “Agentic AI is the next frontier of AI development, and delivering on this opportunity requires full-stack optimization across a system of LLMs to deliver efficient, accurate AI agents.” Meta’s collaboration with Nvidia seeks to harness open models to expedite the creation of custom AI agents for businesses.
Several leading providers of AI platforms, including SAP and ServiceNow, are anticipated to be the early adopters of the Llama Nemotron models. SAP's Chief AI Officer, Philipp Herzig, highlighted the transformative potential of AI agents in enterprises, asserting, “AI agents that collaborate to solve complex tasks across multiple lines of the business will unlock a whole new level of enterprise productivity beyond today’s generative AI scenarios.” Meanwhile, Jeremy Barnes, vice president of platform AI at ServiceNow, remarked, “The improved performance and accuracy of Nvidia’s open Llama Nemotron models can help build advanced AI agent services that solve complex problems across functions, in any industry.”
The newly announced models will be made accessible as downloadable resources or through Nvidia NIM microservices, facilitating easy deployment across clouds, data centres, PCs, and workstations. Nvidia has designed the Llama Nemotron and Cosmos Nemotron models in three configurations—Nano, Super, and Ultra—to accommodate varying scale requirements for deploying AI agents.
These models can also be customised for specific business needs using Nvidia NeMo, a microservice that simplifies data curation and model evaluation while ensuring that AI responses remain relevant and focused. Additionally, the Nvidia Blueprints for agentic AI will streamline the process of developing bespoke applications using Nvidia’s advanced AI tools.
As part of the ongoing evolution in AI technologies, Llama Nemotron and Cosmos Nemotron models will be available for download and as hosted application programming interfaces (APIs) via the Nvidia platform and Hugging Face. Members of the Nvidia Developer Program will have free access for development, testing, and research purposes.
The advancements unveiled at CES 2025 position Nvidia at the forefront of the AI transformation in business, signalling potential far-reaching changes in how organisations will employ technology to enhance operational efficiency.
Source: Noah Wire Services