NVIDIA's open VLM family expands efficient local and data-center vision-language options.
NVIDIA expands its open Nemotron family with two new vision-language models: Super VL for data-center use and Bolt 74B-A9B, a hybrid model sized to run locally on 24GB.