Nvidia releases 120B-parameter open model for enterprise AI agents

Nvidia has launched Nemotron 3.5 Super VL, a 120-billion-parameter open-source AI model designed for enterprise applications. The model is built to handle tasks common in business workflows, such as extracting data from tables, interpreting charts, and cross-referencing documents and websites.

According to Nvidia, the model outperforms larger frontier models on specialized benchmarks tailored to these enterprise use cases. The release positions Nvidia's open-model strategy alongside its proprietary offerings and reflects growing demand for customizable AI systems that companies can run on their own infrastructure.