Nvidia Launches Open Models for Enterprise and PC Agents
Nvidia expands its AI portfolio with open-weight models designed for enterprise data centers, always-on agents, and local edge computing on new PCs.
Radwan AbdulhadiEditor and technology writer
New Open-Weight Models for Enterprises and PCs
Nvidia has announced a new lineup of open-weight models built for always-on agents and edge computing. The hardware and software giant is targeting enterprise companies that prefer to run their own open models within private data centers, while also pushing private, on-device AI agents for everyday users.
Leading the enterprise tier is the Nemotron 3.5 Super VL, a large 120-billion-parameter model designed for agentic tasks. While featuring fewer parameters than the Nemotron 3 Ultra model, Nvidia reports that it achieves similar accuracy levels in instruction-following and tool-use applications.
Enterprise Performance and Industry Adoption
According to Nvidia, the Super VL model can complete specific agentic tasks faster than competing offerings from OpenAI and Anthropic when post-trained. These specialized workflows encompass document intelligence, financial accounting, scientific experimentation, medical insurance claims, and security attack resilience.
To support these claims, the company highlighted several major corporate users currently leveraging the Super VL model. Notable adopters include Skydance, EY, Ramp, Crowdstrike, Perplexity, and Salesforce.
Local Edge Computing and New AI PCs
Alongside its enterprise offerings, Nvidia introduced two lightweight Bolt models aimed at local execution on hardware like the Nvidia RTX Spark chip and DGX Spark workstations. These machines are being released right now into the market through partners like Microsoft and Dell.
Kari Briski, Nvidia’s vice president of Open Models and Software, emphasized the flexibility provided by the new portfolio. In a blog post discussing the rollout, Briski noted that organizations can post-train Super VL on proprietary data, deploy it locally, and maintain complete ownership over their infrastructure.
Breakdown of the Bolt Model Family
The Bolt lineup addresses the growing consumer demand for privacy-focused AI assistants that operate without transmitting sensitive family, financial, or health information to the cloud. Users can delegate everyday projects locally on their devices.
The larger Nemotron 3.5 Bolt 74B-A9B 24GB utilizes 74 billion parameters to balance speed with multistep reasoning, web browsing, and tool calling. Meanwhile, the smaller Bolt E4B model is tailored specifically for conversational agent applications.
Availability and Market Strategy
Nvidia confirmed that the Super VL and Bolt 74B-A9B models are available immediately, while the compact Bolt E4B is scheduled for release later this month. By addressing both enterprise data centers and local consumer hardware, the GPU maker continues to expand its ecosystem amid competition from alternative chip architectures.
Sources
- TechCrunchNvidia launches new open models to run agents in the enterprise and on PCs
Continue chronologically
Related entity coverage
Related stories
Nvidia RTX Spark Laptops Launch With Devices From Major Brands
Nvidia RTX Spark laptops are officially on sale from brands like Microsoft, Asus, HP, Lenovo, and Dell, with prices starting at $2,599.
Microsoft Launches Surface Laptop Ultra with Nvidia RTX Spark
Microsoft unveils the Surface Laptop Ultra and Dev Box, powered by Nvidia RTX Spark chips and a revamped Windows 11 featuring Execution Containers.
Nvidia RTX Spark Laptops Launch With Prices Reaching $7,000
Discover the pricing and specifications of the new Nvidia RTX Spark laptops from Dell, Lenovo, MSI, HP, and Asus, launching as early as October 16th.
Microsoft Pivots Windows 11 to Build an Agentic OS
Microsoft is transforming Windows 11 into an agentic operating system, introducing hybrid intelligence, local AI models, and new hardware at a recent event.
