What is NVIDIA Nemotron

NVIDIA Nemotron is a family of open, customizable AI models designed to accelerate the development and deployment of enterprise AI agents. Built for reasoning, multimodal understanding, and agentic workflows, the Nemotron 3.5 family further expands these capabilities with enterprise knowledge retrieval and speech intelligence, enabling AI agents to process complex information, access relevant knowledge, and execute multistep tasks.

With flexible deployment options across on-premises data centers and cloud environments, Nemotron helps enterprises build specialized AI applications while maintaining greater control over their models, data, and infrastructure.

What are the key capabilities of NVIDIA Nemotron 3.5?

Multimodal Agentic AI — Nemotron 3.5 Super VL

NVIDIA Nemotron 3.5 Super VL is an open, customizable vision-language model designed for specialized agentic workflows. It combines reasoning, tool use, and multimodal understanding in a single model, allowing AI agents to process text, images, documents, and video.

It supports enterprise applications such as document intelligence, video understanding, customer service automation, and coding assistance, helping organizations develop AI agents that can interpret complex information and complete multistep tasks.

Enterprise Knowledge Retrieval — Nemotron 3.5 Retriever

NVIDIA Nemotron 3.5 Retriever provides a collection of specialized models designed to improve how enterprise AI agents access and use information across documents, images, tables, and other data sources.

It consists of three key components:

Nemotron 3.5 Parse: Extracts and converts complex enterprise documents, including PDFs, charts, and tables, into structured, AI-ready content.
Nemotron 3.5 Embed: Enables multimodal search and retrieval across text, images, and enterprise documents.
Nemotron 3.5 Rerank: Prioritizes the most relevant retrieved information to improve response accuracy and reduce unnecessary context.

Together, these capabilities strengthen retrieval-augmented generation (RAG) and enterprise knowledge workflows, helping AI agents retrieve more relevant information and generate more reliable responses.

Enterprise Speech AI — Nemotron 3.5 Transcribe

NVIDIA Nemotron 3.5 Transcribe delivers highly accurate, low-latency enterprise speech recognition and transcription across diverse acoustic environments. It enables AI agents to process multi-speaker audio, meetings, and call center interactions into high-fidelity text for downstream reasoning and RAG workflows.

Enterprise agentic AI applications. Combining multimodal understanding, enterprise knowledge retrieval, and speech intelligence, Nemotron 3.5 enables organizations to develop AI agents capable of understanding complex information, accessing relevant enterprise knowledge, and supporting multistep workflows.

With flexible deployment options, NVIDIA Nemotron 3.5 helps enterprises build specialized AI applications while maintaining greater control over their models, data, and infrastructure.

Why do enterprises need NVIDIA Nemotron?

As enterprises move beyond traditional AI chatbots toward autonomous AI agents, they require models capable of understanding complex information, retrieving relevant knowledge, and executing multistep tasks.

Enterprise AI adoption also introduces challenges related to data privacy, inference efficiency, scalability, and deployment flexibility. NVIDIA Nemotron helps organizations address these needs by providing customizable AI models that can be integrated into enterprise workflows and deployed on infrastructure aligned with their performance and data-governance requirements.

Key benefits of NVIDIA Nemotron

Agentic AI capabilities: Supports reasoning, tool use, and multistep workflows to help AI agents perform complex enterprise tasks.
Multimodal understanding: Enables AI applications to process diverse information, including text, documents, images, and other supported data types.
Enterprise knowledge retrieval: Supports AI workflows that retrieve and use relevant information from enterprise knowledge sources to improve response accuracy.
Customization and flexibility: Open models allow organizations to adapt AI capabilities to industry-specific applications and deployment requirements.
Scalable AI inference: Enables efficient AI model deployment across supported GPU-accelerated infrastructure, from individual servers to large-scale AI systems.

How does ASUS enable NVIDIA Nemotron-powered AI?

ASUS delivers high-performance AI infrastructure designed to support the development, deployment, and scaling of enterprise AI applications powered by NVIDIA Nemotron. From standard enterprise servers to high-density GPU platforms and massive rack-scale AI PODs, ASUS provides optimized computing architectures for every stage of agentic AI adoption:

ASUS GPU servers powered by NVIDIA RTX PRO™ 6000 Blackwell Server Edition: ASUS ESC8000A-E13P and ESC8000A-E13X support up to eight NVIDIA RTX PRO™ 6000 Blackwell Server Edition GPUs, offering flexible GPU acceleration for enterprise AI inference, multimodal applications, and agentic AI workloads.
ASUS AI POD with NVIDIA GB300 NVL72: Designed for large-scale AI infrastructure, ASUS AI POD integrates NVIDIA Blackwell Ultra GPUs, NVIDIA Grace CPUs, high-speed networking, and liquid cooling into a rack-scale platform. It provides the computing foundation for demanding AI inference, advanced reasoning, and large-scale agentic AI deployments.
ASUS AI POD with NVIDIA Vera Rubin NVL72: Built for next-generation AI factories, the ASUS XA VR721-E3 integrates 72 NVIDIA Rubin GPUs, 36 NVIDIA Vera CPUs, high-bandwidth networking, and a fully liquid-cooled rack-scale architecture to support massive AI workloads, advanced reasoning, and high-throughput inference.

By offering scalable infrastructure from GPU servers to integrated AI POD solutions, ASUS helps enterprises build AI environments that align with their performance, scalability, and data-control requirements.