[CLSTR]
< Back to situations

Monitor this situation.

[SITUATION] · [QUIET] · [TECHNOLOGY]

3 clusters · 21 sources · 7 days · First seen · Last updated

Nvidia Nemotron AI model and software development

Overview

Nvidia is expanding its presence in the AI software and open-weight model market through the development and release of its Nemotron model family.

Reports indicate the company is working on Nemotron 4, a next-generation series that may include a version with at least one trillion parameters. This development follows significant financial commitments, with Nvidia reportedly tripling its cloud spending on in-house model training to $28 billion through 2031.

To optimize AI agent efficiency, Nvidia released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts (MoE) model. By utilizing proprietary NVFP4 4-bit floating-point precision and Quantization-Aware Distillation (QAD), the model reduces memory requirements from 66 GB to 22 GB and achieves up to four times the throughput of full-precision versions.

Accompanying this release is NeMo Switchyard, an open-source routing library designed to direct requests to the most suitable models based on latency, cost, and quality. Nvidia claims this router can achieve a “74% cost reduction” compared to using frontier models exclusively, though it may result in a “6% reduction in accuracy.” Implementation challenges have been noted, including potential latency of approximately 700 milliseconds per request and the fact that the judging model used for routing can consume up to 21.2% of total costs.

Entities

OpenAI · Nvidia · NeMo Switchyard · DeepSeek · LangChain

Timeline

  1. 24 days ago

    [TECHNOLOGY] 2 sources
    Nvidia launches NeMo Switchyard to optimize AI model costs

    Nvidia launched NeMo Switchyard, an open-source router that optimizes AI costs by directing requests to the most efficient language model for each specific task.

  2. 25 days ago

    [TECHNOLOGY] 10 sources
    NVIDIA launches Nemotron 3.5 Lightning to boost AI agent efficiency

    NVIDIA launches Nemotron 3.5 Lightning, an efficient MoE model for AI agents, while companies like Advantech and Hocheon Technology integrate AI and robotics into industrial and manufacturing sectors.

  3. about 1 month ago

    [TECHNOLOGY] 9 sources
    Nvidia developing Nemotron 4 AI model with one trillion parameters

    Nvidia is developing Nemotron 4, a new AI model family that may feature a version with at least one trillion parameters to compete in the open-weight model market.

Sources

bitcoinethereumnews.com · borncity.com · brasilemfolhas.com.br · breezyscroll.com · businesstoday.me · developer.nvidia.com · epochtimes.com · equitypandit.com · finance.technews.tw · gamemag.it · itdaily.be · muycomputerpro.com · news.m.pchome.com.tw · news.yesky.com · nextapple.com.sg · pcauthority.com.au · playtennis.com · techgenyz.com · the-decoder.com · tuttotech.net · tw.nextapple.com

This summary has been updated 3 times: see revision history