NVIDIA Releases Nemotron 3.5 Super VL

NVIDIA has expanded its Nemotron family with the release of Nemotron 3.5 Super VL, an open 120-billion-parameter model built for specialized always-on agents. The company says the model delivers the fastest task completion and highest accuracy in its class, targeting developers who need autonomous systems that can operate continuously without constant human intervention. The open model is designed to give teams more control over deployment, fine-tuning, and integration. Alongside Super VL, NVIDIA introduced edge-focused options, including Nemotron 3.5 Bolt 74B-A9B 24GB. That model is described as balancing intelligence and speed per gigabyte, a critical consideration for devices with limited memory and power. NVIDIA also introduced another edge model, further extending the Nemotron line beyond data centers. The move shows how AI development is spreading from massive cloud clusters to desktops, workstations, and local devices. The strategy matters because always-on agents need to run reliably in many environments. A data-center model may offer maximum capability, but an edge model can reduce latency, improve privacy, and lower costs by processing data locally. For developers building assistants, robotics, monitoring tools, or enterprise automation, having open models at different scales makes it easier to match the model to the task. By expanding Nemotron from data center to desktop, NVIDIA is positioning itself as a supplier of both hardware and open model software. The release also adds pressure to the broader open-weight AI market, where efficiency, licensing, and deployment flexibility are increasingly important. If Super VL and the new edge models live up to their performance claims, they could become practical choices for teams building the next generation of autonomous agents.