New Alchemy Studio
built as a mosaicSign in
← Catalog/NVIDIA (Nemotron)

Nemotron

NVIDIA (Nemotron)United StatesOpen sourceOpen weightsnvidia-opentext

Open NVIDIA-trained models (often Llama-based).

Release history (3)

  1. Llama-Nemotron Ultra
    stableApr 7, 2025
    nvidia/Llama-3.3-Nemotron-Super-49B-v1253B params

    Reasoning Ultra.

  2. Llama-3.1-Nemotron-70B
    stableOct 15, 2024
    nvidia/Llama-3.1-Nemotron-70B-Instruct-HF70B params

    RLHF on Llama-3.1.

  3. Nemotron-4 340B
    stableJun 14, 2024
    nvidia/Nemotron-4-340B-Instruct340B params

    340B trained on synthetic data.

Spotted something out of date? Edit in admin.