NVIDIA has expanded its Nemotron 3.5 artificial intelligence model series by introducing Nemotron 3.5 Super VL, a large-scale visual language model designed for AI agents. This model boasts 120 billion parameters and can achieve up to 5 times the throughput of similarly sized models when performing agent tasks on a single GPU. Meanwhile, NVIDIA has released two lightweight models: Nemotron 3.5 Bolt 74B-A9B (24GB) and Nemotron 3.5 Bolt E4B (4GB), tailored for scenarios such as local hardware deployment and local voice assistants, respectively. This release enhances NVIDIA's AI deployment ecosystem, spanning from data centers to end-user devices.
