Nvidia open models let firms run agents on-site or on a PC
In brief
Nvidia has launched open-weight models for AI agents, according to the company. The 120-billion-parameter Nemotron 3.5 Super VL targets enterprise use, while two lighter Bolt models are designed to run on new AI PCs and workstations with Nvidia chips, sold through partners including Microsoft and Dell. Nvidia says Super VL and the larger Bolt model are available now, with the smallest Bolt model due later this month.
Source : TechCrunch - AI - « Nvidia launches new open models to run agents in the enterprise and on PCs »
What it changes for an SME
Most SMEs of 10 to 250 people will not run the 120-billion-parameter Super VL model without dedicated Nvidia infrastructure. The smaller Bolt models target new AI PCs and workstations sold through partners such as Microsoft and Dell, so some agent work could stay on a local machine. According to Nvidia, you can post-train Super VL on your own data and deploy it on your own systems, which may cut cloud transfers and related GDPR risk. Hardware cost and operating skills are still required; Nvidia does not give figures for time or money saved.
What you can do
- Check whether planned PCs or workstations use Nvidia RTX Spark or DGX Spark chips before assuming local deployment is possible.
- If you use AI on client or staff data, compare a local open model with cloud APIs for GDPR and the EU AI Act.
- Note Nvidia's stated availability: Super VL and the 74-billion-parameter Bolt model now, the smallest Bolt model later this month.
Let's talk about the time you could save.
In 20 minutes, we find where AI could save you time or money. Free, no commitment.