Nvidia has introduced a new set of open-weight AI models designed for enterprise agents and computers that can run AI workloads locally. The lineup includes the large Nemotron 3.5 Super VL and two smaller Bolt models.
A model for enterprise workflows
With 120 billion parameters, Super VL is built for business agents that handle tasks such as document analysis, accounting and scientific work. Nvidia says it can match the instruction-following and tool-use accuracy of its larger Nemotron 3 Ultra model while using fewer parameters. The company also reports faster performance than selected competing models on specialized agent tasks.
Nvidia says organizations can adapt Super VL using their own data and run it on their own infrastructure. The model is available now, and the company says businesses including EY, Salesforce and Perplexity are using it.
Smaller models for on-device AI
The Bolt models are intended for local use on Nvidia-powered AI PCs and DGX Spark workstations. The 74-billion-parameter Bolt 74B-A9B targets multistep reasoning, tool use and web browsing. The smaller Bolt E4B is designed for conversational agents and is expected later this month. Nvidia says local processing can let users delegate everyday tasks without sending them to cloud services.
By offering models for both data centers and personal computers, Nvidia is expanding options for organizations and users seeking more control over where AI tasks run. Wider access to capable on-device models could shape a future of AI assistance that is more customizable and locally managed.
