Nvidia has launched a new set of open-weight models built for enterprise agents, edge computing and local AI workloads.
The flagship Nemotron 3.5 Super VL has 120 billion parameters and is designed for enterprise use cases, including tool use and instruction following. Nvidia said it delivers accuracy comparable to its larger Nemotron 3 Ultra model while using fewer parameters.
The company also introduced two smaller Bolt models for local deployment on PCs powered by Nvidia’s RTX Spark chip and DGX Spark workstations. The 74-billion-parameter Nemotron 3.5 Bolt 74B-A9B 24GB is designed for multistep reasoning, tool calling and web browsing, while the smaller Bolt E4B targets conversational agents.
Super VL and Bolt 74B-A9B are available now, with Bolt E4B expected later this month. Nvidia said the models can run on an organization’s own infrastructure and be post-trained on its data. Local deployment also allows users to run workloads without sending them to the cloud.
Nvidia said companies including Skydance, EY, Ramp, CrowdStrike, Perplexity and Salesforce are using Super VL. It also claims the model can outperform competing systems from Anthropic and OpenAI on selected agentic tasks after post-training.
Source: TechCrunch
