Nvidia Launches Open AI Models for Enterprise Agents and Local PCs
Nvidia has added to its software ecosystem by releasing open-weight artificial intelligence models for autonomous agent operation on both enterprise servers and local PC’s. The offerings include the 120 billion-parameter Nemotron 3.5 Super VL model for enterprise processes and two compact Bolt versions meant for on-device use.
Through the provision of open weights, Nvidia seeks to provide users with options for highly performing AI deployments which are independent of costly cloud resources and are capable of maintaining user privacy.
The flagship model of the Nemotron 3.5 Super VL series is targeted at enterprise processes like document intelligence, financial accounting, and automated customer processes, and deals with multi-step reasoning and tool invocations.
The compact Nemotron 3.5 Bolt series consists of models like the 74 billion-parameter Bolt 74B-A9B and Bolt E4B, which can be used for running private assistants on local AI PC’s and workstations with Nvidia RTX Spark processors. Operationally, making open-weight models public ensures that corporate IT departments can keep tight control over proprietary software and any data.
Enterprise software companies such as Salesforce are already integrating Nemotron architectures into special workflows of automated operations. Thus, by providing competent open-weight solutions, Nvidia saves the cost of running inference from proprietary cloud API subscriptions and makes it possible to scale the number of digital workers using existing company hardware infrastructure.
The move shows the current industry trend towards localized computing and open-source models due to the need for businesses to have alternative solutions other than the cloud. By linking its software models to GPU hardware, Nvidia is able to maintain its dominance in the market within both data center and emerging AI PC segments.
The introduction of Nvidia’s open software models allows creating a framework for enterprise automation and promotes the adoption of secure, private and localized AI agent operation at corporate enterprises.
By enabling post-training and hosting of the personalized agentic software inside of the company, Nvidia saves its expenses on operations while increasing its hardware lock-in in enterprise AI hardware infrastructure.
