Nemotron 3.5 Super VL

NVIDIA Nemotron 3.5 Super VL is an open, customizable multimodal model designed for specialized digital agents. Key capabilities include: • Integrates text, image, document, and video input • Incorporates reasoning and tool use within a single model • Achieves significantly higher agentic throughput on a single GPU • Offers customizable checkpoints, training recipes, and datasets This 120B mixture-of-experts model, with 12B active parameters, processes diverse data types to facilitate complex tasks. It empowers developers to build sophisticated agents capable of handling long, intricate enterprise workflows, such as reading documents, cross-referencing information, and adhering to company policies. Nemotron 3.5 Super VL is distributed via Hugging Face, including training resources and a NIM container, making it accessible for developers and researchers. It is an ideal platform for creating highly specialized agents for various industries, enabling advanced computational processes and efficient handling of multimodal data streams. This model enhances the capabilities of specialized agents, allowing them to perform intricate operations with greater speed and accuracy.