NVIDIA's Nemotron 3.5 Super VL: 5x faster, one-fifth the size

NVIDIA is positioning the Nemotron family as a tiered lineup rather than a single flagship. Lightning, released in August, handles high-throughput workloads; Ultra handles the hardest reasoning tasks; and Bolt brings intelligence to the edge. Super VL slots into the middle as the high-volume agentic workhorse, and NVIDIA says the three open models together cover deployments from large enterprise servers down to edge devices and local hardware.

The customer list NVIDIA assembled around Super VL shows where it expects the model to earn its keep. Scale AI post-trained it for Skydance's vendor credit finance agent and EY's M&A due diligence agent, AfterQuery customized it for Hebbia's credit agreement analysis, and AI21 and Alice worked on cybersecurity attack resilience. Platform companies including Atlassian, CrowdStrike, Dassault Systèmes, Perplexity, Salesforce and Synopsys are also integrating or customizing it.

NVIDIA ties that uptake to the model's openness. Super VL ships with open weights, open datasets and training recipes, and NVIDIA is releasing open datasets on Hugging Face spanning agentic tool use, finance, instruction following, long context, fairness and vision-language understanding. For buyers weighing open against closed frontier models, the pitch is control over where the model runs and what data it sees, rather than raw benchmark leadership alone.