GitSeonjeap
English

← News

Local AI

NVIDIA Releases Nemotron for Science for Scientific Agents

Posted As of

NVIDIA announced a package based on Nemotron 3.5 Super VL. It includes a chemistry dataset and a post-training recipe, and the IR copy states that the model is optimized to run locally on a 128GB DGX Spark.

What happened

  • According to the NVIDIA Newsroom, it was announced at SC26 (Chicago) on 2026-11-16.
  • The package is based on the Nemotron 3.5 Super VL open model.
  • It comes with a chemistry dataset and a post-training recipe for building scientific agents.
  • According to the IR copy, the model is optimized to run locally on a 128GB NVIDIA DGX Spark (official).
  • LitmusBench v0.2 consists of about 50,000 questions (official).
  • The rate at which scientists preferred the explanations over the base model is 96% (official). The evaluation method has not been confirmed.
  • Parameter count, license, and weight size are not listed in the notes.

Our take

The fact that an open model, data, and recipe are released together is meaningful for users interested in local execution. However, the license and weight size have not been confirmed, and there are no results for running it on Macs or PCs other than the DGX Spark, so the actual conditions of use remain unclear.

Sources and related posts

More in Local AI

  1. NVIDIA RTX Spark Compact Desktops Begin Shipping (blogs.nvidia.com)

    Compact RTX Spark desktops from ASUS, Dell, and MSI have launched as of the announcement date and are positioned for local AI agents and model fine-tuning. Pricing was not confirmed on NVIDIA's page.

    Local AI| |1 sources |Our take included
  2. llama.cpp b12497 adds tiny-voca support (github.com)

    llama.cpp pre-release b12497 adds tiny-voca support to the multimodal path and improves G2P conversion. The release page has no performance figures or description of the target model.

    Local AI| |1 sources |Our take included
  3. llama.cpp b12441 Pre-release Fixes WebGPU Mask Bug (github.com)

    llama.cpp pre-release b12441 (2026-11-16) fixed a bug where the WebGPU flash-attention shader read the mask without its head dimension. b12436, released the same day, fixed a Vulkan 32-bit compilation issue.

    Local AI| |1 sources |Our take included

All news