Nvidia announces new additions to its Nemotron large language model family aimed at running “agentic” AI systems across both servers and edge devices. The company introduces Nemotron 3.5 Super LV, positioned for always-on agents, and adds an edge-focused option, Nemotron 3.5 Bolt, targeted at local deployments.
Across outlets, the emphasis is on efficiency and deployment flexibility. One report highlights a 120B vision-language variant, Nemotron 3.5 Super VL, describing capabilities for multimodal use cases. Another outlet focuses on the ability to run a 74B agent model within 24GB of memory, framing the update as a step toward bringing larger models to consumer or laptop-class hardware. Nvidia presents the lineup as covering the spectrum from large enterprise systems to local and edge environments, with customization options for different deployment needs.
