Key Takeaways
- Nvidia has launched Nemotron 3.5 models for local AI agents and enterprise workloads.
- The Super VL model is designed for enterprise use, while smaller Bolt models target local PCs and workstations.
- Companies like Skydance, EY, and Salesforce are already using the Super VL model.
Nvidia has introduced new open-source AI models, the Nemotron 3.5, designed for local deployment and enterprise applications. These models include the Super VL, which is optimized for enterprise agents, and two smaller Bolt models, aimed at local AI agents on Nvidia-powered PCs and workstations.
The Nemotron 3.5 Super VL, with 120 billion parameters, is positioned as a powerful tool for instruction-following and tool-use tasks, offering similar accuracy to competing models like Claude Opus 5 and GPT-6.1 Sol. It is faster in completing certain tasks, according to Nvidia.
Nvidia claims that the Super VL can run specific workloads such as document intelligence, financial accounting, and medical insurance claims more efficiently, while also being resilient against security attacks.
Several companies, including Skydance, EY, Ramp, CrowdStrike, Perplexity, and Salesforce, are already utilizing the Super VL model, indicating its potential for widespread adoption.
The smaller Bolt models, the Nemotron 3.5 Bolt 74B-A9B 24GB and the Bolt E4B, are designed for local agentic workloads and conversational AI agents, respectively. These models are intended for new AI PCs powered by Nvidia’s RTX Spark chip and workstations using DGX Spark.
Nvidia emphasizes the benefits of running AI agents locally, allowing users to handle everyday tasks without sending sensitive data to the cloud. This feature is particularly useful for tasks involving health, finances, or family information.
The new releases expand Nvidia’s existing Nemotron family, which covers multimodal AI, reasoning, coding, mathematics, and enterprise automation. This move comes as major AI companies explore hardware beyond Nvidia, including AWS Trainium and Cerebras, while also developing their own chips.
Nvidia is targeting growing demand from companies that want to operate open models inside their own data centers. The Nemotron 3.5 Super VL and Bolt 74B-A9B are now available, with the Bolt E4B expected to arrive later this month.
