Nvidia introduced three new open models under its Nemotron family, aimed at spanning the full spectrum of enterprise and edge AI deployments. The flagship release, Nemotron 3.5 Super VL, is a 120-billion-parameter multimodal model designed to handle complex visual and textual reasoning tasks. The lineup also features two smaller variants, the Nemotron 3.5 Bolt 74B-A9B and Bolt E4B, tailored for localized hardware, coding, and device-based agents. Built to integrate with enterprise data and post-training practices, Super VL has shown competitive performance on specialized task benchmarks. The models are part of Nvidia’s broader push to support multi-model strategies, offering flexible deployment across cloud data centers and local edge infrastructure.

Full report : NVIDIA releases the 120-billion-parameter Nemotron 3.5 Super VL alongside two lightweight Bolt models, pushing advanced visual reasoning and local edge capabilities into enterprise workflows.