Nvidia Expands Nemotron Family with Highest Efficiency Super VL and Bolt Models (blogs.nvidia.com)

NVIDIA has unveiled its latest additions to the Nemotron family, introducing the Nemotron 3.5 Super VL model with a remarkable 120 billion parameters, alongside two new edge models: the Nemotron 3.5 Bolt 74B-A9B and Bolt E4B. The Super VL model stands out as the fastest open agentic model in its class, offering exceptional accuracy in multimodal tasks such as instruction following and visual understanding, while operating at roughly one-fifth the parameter size compared to larger models. With its release, organizations can now choose an optimized model tailored to their deployment needs—ranging from large servers to local edge devices—enhancing their ability to operate sophisticated AI agents without sacrificing performance. The significance of these releases is underscored by the model's adaptability; organizations can post-train Super VL on their proprietary data for specialized tasks, allowing for a competitive edge in sectors such as finance, cybersecurity, and healthcare. The models facilitate local task execution, ensuring that sensitive data remains secure and private. With built-in support for seamless integration across NVIDIA's hardware stack and open datasets for further customization, the Nemotron 3.5 models signal a major step forward in empowering businesses to harness AI's potential effectively and efficiently.