January 6, 2025 by Kari Briski
In a significant advancement for artificial intelligence, NVIDIA has unveiled the Nemotron Model Families, designed to revolutionize the emerging domain of agentic AI. These state-of-the-art models aim to optimize and empower AI agents, enhancing productivity and capability across various industries. Building on NVIDIA’s legacy of innovation in AI and computing technologies, the Nemotron models are set to provide pivotal support for businesses looking to leverage intelligent automation.
As part of its latest AI solutions, NVIDIA has introduced two main model categories: the Llama Nemotron large language models (LLMs) and Cosmos Nemotron vision language models (VLMs). Available as NVIDIA NIM microservices, these models are designed to seamlessly integrate with accelerated systems, thereby enhancing the functionality and efficiency of AI agents.
The Llama Nemotron models, built upon the widely acclaimed Llama foundation models, are engineered to deliver optimal efficiency and accuracy for enterprise-focused AI agents. They serve as foundational building blocks capable of handling diverse applications such as customer support, fraud detection, and supply chain management. By optimizing compute efficiency and accuracy, these models empower enterprises to develop custom AI agents tailored to specific operational needs.
“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,” said Ahmad Al-Dahle, vice president and head of GenAI at Meta.
The Cosmos Nemotron VLMs further complement the capabilities of AI agents, enabling them to analyze and interpret complex visual data. These models support developers in creating agents that can process images and video content from varied environments, including autonomous vehicles, healthcare facilities, and retail spaces. This integration of visual and linguistic understanding is crucial for AI applications in fields demanding perceptual acumen and swift decision-making.
NVIDIA’s commitment to providing versatile AI tools is evident in its NeMo microservices, which facilitate the customization and deployment of AI models at multiple scales. Businesses can choose from Nano, Super, and Ultra models, optimizing for factors like cost-efficiency, accuracy, and performance. Additionally, NVIDIA offers the NeMo Retriever for integrating enterprise data, enhancing model alignment with specific business processes.
“AI agents that collaborate to solve complex tasks across multiple business functions will unlock unprecedented productivity,” said Philipp Herzig, chief AI officer at SAP.
Enterprises are encouraged to utilize NVIDIA Blueprints for agentic AI, which leverage the power of Cosmos Nemotron, Llama Nemotron, and NeMo Retriever to accelerate app development. NVIDIA’s strategic partnerships with leading platforms like SAP and ServiceNow underscore the transformative potential of these models in redefining business operations.
Developers and enterprises eager to explore NVIDIA’s cutting-edge AI solutions can access the Nemotron models via build.nvidia.com and Hugging Face. These models are also available for download and use as hosted APIs, with no charge for members of the NVIDIA Developer Program. Enterprise deployment is facilitated through the NVIDIA AI Enterprise software platform, ensuring robust support and integration capabilities.
For those interested in staying abreast of NVIDIA's ongoing innovations in AI and automation, registrations for updates and participation in upcoming events like CES are encouraged.
```
