Nvidia launches new open models to run agents in the enterprise and on PCs

Your AI agents can now run locally on your PC, handling tasks without sending data to the cloud.

Nvidia has launched new open-weight models, Nemotron 3.5 Super VL and two Bolt models, aimed at enabling always-on agents for both enterprise and personal computing. The Nemotron 3.5 Super VL is a large, 120-billion-parameter model designed for enterprise agents, offering high accuracy in instruction-following and tool-use tasks. Nvidia claims it outperforms competing models from Anthropic and OpenAI in agentic tasks like document intelligence and financial accounting, with early adopters including Skydance, EY, and Salesforce. Complementing this, the lightweight Bolt models are intended for local execution on new AI PCs powered by Nvidia's RTX Spark chip and DGX Spark workstations, allowing users to delegate tasks without cloud processing. The Bolt 74B-A9B model focuses on local agentic tasks balancing speed and reasoning, while the smaller Bolt E4B is designed for conversational agents. Super VL and Bolt 74B-A9B are available now, with Bolt E4B expected later this month.

Nvidia's release of open-weight models, particularly Nemotron 3.5 Super VL, signals a strategic push to empower enterprises with greater control over their AI deployments. By offering models that can be post-trained on proprietary data and run on in-house infrastructure, Nvidia caters to the growing demand for data privacy and customization in enterprise AI. This move challenges the dominance of closed, proprietary models and positions Nvidia as a key enabler for organizations seeking to build and own their AI agentic capabilities, potentially reducing reliance on hyperscale cloud providers for core AI functions.

The introduction of the local Bolt models for AI PCs indicates Nvidia's ambition to democratize AI agent capabilities for consumers and knowledge workers, directly competing with the nascent market for AI assistants. By enabling these agents to run on-device, Nvidia addresses consumer privacy concerns and creates a new hardware market for its chips, facilitating local processing of sensitive data. This strategy could foster a new generation of personal productivity tools and privacy-focused AI companions, shifting how individuals interact with and delegate tasks to their personal computing devices.