[CLSTR]
< Back to situations

Monitor this situation.

[SITUATION] · [ACTIVE] · [TECHNOLOGY]

2 clusters · 5 sources · 8 days · First seen · Last updated

NVIDIA Nemotron 3.5 model releases

Overview

NVIDIA has expanded its Nemotron 3.5 model lineup with releases targeting different hardware capabilities. The Nemotron 3.5 Super VL, a 120B vision-language Mixture-of-Experts model, is designed for high-memory systems, requiring approximately 128GB of memory for local execution. This model features a Mamba-2/Transformer hybrid architecture and a vision encoder capable of processing text, images, and video.

Following this, NVIDIA released the Nemotron 3.5 Bolt E4B, a smaller model optimized for low-memory environments such as voice assistants and robotic interfaces. Derived from Google’s Gemma 4 E4B-it base, the Bolt E4B utilizes reinforcement learning to improve instruction following and tool calling. It is designed to operate with approximately 4GB of memory usage.

Entities

Nvidia · Nemotron 3.5 Bolt E4B · Google · Nemotron 3.5 Super VL · Gemma 4

Timeline

  1. [TECHNOLOGY] 3 sources
    NVIDIA releases Nemotron 3.5 Bolt E4B small language model

    NVIDIA released Nemotron 3.5 Bolt E4B, a small language model optimized for local agents, voice assistants, and robotics with a target memory footprint of approximately 4GB.

  2. [TECHNOLOGY] 2 sources
    Nemotron 3.5 Super VL model can run locally on 128GB memory systems

    NVIDIA's Nemotron 3.5 Super VL vision-language model can run locally on hardware with 128GB of memory using community-converted formats, bypassing the need for specialized Blackwell GPUs.

Sources

dev.classmethod.jp · dev.to · donanimhaber.com · hackernews.com · teknoblog.com