Nvidia squeezed a 74B agent model into 24GB so it can run on your laptop
Nvidia on Thursday announced a few new additions to its Nemotron family: Nemotron 3.5 Super VL, a 120B vision-language model for The post Nvidia squeezed a 74B agent model into 24GB so it can run on your laptop appeared first on The New Stack .
Nvidia has unveiled three new models in its Nemotron family, which can run on laptops and deliver efficient performance for various tasks. The largest model, Nemotron 3.5 Super VL, is a 120-billion-parameter vision-language model designed for enterprise agents. It combines reasoning and tool use with image and document understanding, offering comparable accuracy on multimodal tasks to Nvidia's largest reasoning model at only one-fifth the parameter count.
The new Bolt models, Nemotron 3.5 Bolt 74B-A9B 24GB and Nemotron 3.5 Bolt E4B 4GB, are smaller and designed for coding, research, and conversational workloads. The 74B-A9B 24GB model, which can run on a laptop with just 24GB of memory, is particularly notable. It uses a hybrid Mamba-2, attention, and mixture-of-experts architecture, and applies two-bit quantization to reduce its memory footprint to 24GB.
This allows the model to achieve impressive performance, scoring 51.0 on Terminal-Bench 2.1, outperforming other comparable models like Qwen3.6-35B-A3B and Gemma4 26B-A4B. Overall, Nvidia emphasizes the speed and efficiency of these models to enable everyday agentic tasks on laptops.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
