A broad open-model portfolio

As part of its sovereign AI push in Europe, NVIDIA expanded its open model family across language, robotics, biology and quantum computing. The company says these models give organizations a foundation they can customize with local data and deploy on infrastructure they control.

The new models

  • Nemotron 3.5 Super VL and Nemotron 3.5 Bolt: Language models that support text and images, helping agents orchestrate workflows and complete specialized tasks at lower cost.
  • Nemotron Retriever: Models for embedding, reranking and document parsing.
  • Kumo: Open-weight foundation models and libraries for forecasting, fraud-risk scoring and recommendations on structured business data.
  • Cosmos 3.5: World models for robots, autonomous vehicles and vision agents.
  • Alpamayo 2 Edge for autonomous vehicles, BioNeMo for healthcare and scientific discovery, and Ising for quantum computing.

Who is using them

According to NVIDIA, the Nemotron models are built with data and research from the Nemotron Coalition, which includes European partners Black Forest Labs, H-Company and Mistral. Companies including Dassault Systèmes, SAP, Salesforce, Shopify, Synopsys and WPP, along with Accenture and EY, are working with the models to build custom models and domain-specific agents. Early Kumo users include Cognite, Coinbase, Databricks, DoorDash, Revolut and Tubi.

Why it matters

Open-weight models let organizations keep data on their own infrastructure and adapt models to local languages. That positions NVIDIA in both the model layer and the hardware that runs them.