Nvidia releases Nemotron Super VL and Bolt models for server and local AI agents
Nvidia released two open Nemotron models for AI agents and scheduled a smaller model for October 23.
NVIDIA
released Nemotron 3.5 Super VL and Bolt 74B-A9B on October 15. The models target AI agents running on enterprise infrastructure and local computers.
Super VL has 120 billion parameters and combines visual understanding with text processing. Nvidia said it can interpret documents, charts and webpages while handling tasks that involve multiple steps. The company claims throughput up to five times that of competing models in its class.
Bolt 74B-A9B uses 2-bit quantization to fit into 24GB of memory. It is designed for local work such as web research and coding. The releases expand a family that also includes Lightning for high-volume workloads and Ultra for difficult reasoning tasks.
Developers can obtain both models now, and Nvidia provides Super VL through its NIM deployment service. The smaller Bolt E4B 4GB is scheduled for October 23, targeting applications including conversational assistants and embedded interfaces.
The day's biggest AI news in one short email, every morning. Plus an instant heads-up when a reset or a big release lands.
More on NVIDIA
- Oct 15Nvidia highlights Nordic AI expansion and Sweden’s shared supercomputerOfficial
- Oct 14Nvidia pledges $250 million in computing access for UK AI safety researchNews
- Oct 14European cities test and deploy Nvidia AI for buildings and public servicesOfficial
- Oct 13Nvidia plans $2 billion in European investments as it backs French startupsNews
- Oct 13Nvidia details software updates ahead of RTX Spark's October 16 launchOfficial
- Oct 12Nvidia shares Vera Rubin rack designs through Open Compute ProjectOfficial