NVIDIA Widens Its Open Model Lineup for Europe, From Agents to Quantum

A broad portfolio built for customization

According to NVIDIA, the open models give organizations a foundation they can customize with local data and deploy on infrastructure they control. The new portfolio includes the NVIDIA Nemotron 3.5 Super VL and Nemotron 3.5 Bolt models, new Nemotron Retriever models for embedding, reranking and document parsing, the new NVIDIA Kumo family of open weight foundation models and libraries for structured business data, and NVIDIA Cosmos 3.5 world models. It also covers NVIDIA Alpamayo 2 Edge for autonomous vehicles, NVIDIA BioNeMo for healthcare and scientific discovery, and NVIDIA Ising for quantum computing.

NVIDIA says Nemotron brings open language models, datasets and recipes to developers building agents. The models support text and images and help agents orchestrate workflows, reason through complex problems and complete specialized tasks at a lower cost. They are built with data, research and expertise from the NVIDIA Nemotron Coalition, which includes European partners Black Forest Labs, H-Company and Mistral.

Dassault Systèmes, Domyn, Dream, Palantir's customers, Salesforce, SAP, Shopify, Synopsys and WPP, along with Accenture and EY, are working with Nemotron models and datasets to create custom models and domain-specific agents. A wider group of post-training partners, including Baseten, CoreWeave, Fireworks, Google Cloud, Nebius and Scale AI, is helping enterprises customize Nemotron for specific workloads.

Kumo targets structured business data

NVIDIA Kumo is a set of open foundation models, libraries and tools that help agents and applications forecast outcomes, score fraud risk and recommend actions from structured business data. It is part of NVIDIA Agent Knowledge Foundation. NVIDIA says Kumo shortens time to prediction and reduces the cost and complexity of building a separate machine learning pipeline for every task.

Developers can start with pretrained models and then fine-tune them on proprietary data when higher performance is critical. The portfolio includes Kumo Structured, with models for relational and tabular prediction, forecasting and anomaly detection, and Kumo Engine, an open library stack for developing and deploying structured data foundation models.

Cognite, Coinbase, Databricks, DoorDash, Revolut and Tubi are among the first to work with Kumo models. Snowflake customers are already using them in industries including advertising, energy, financial services and retail.

Robotics, driving, biology and quantum

NVIDIA Cosmos is a family of open world foundation models for building robots, autonomous vehicles and vision agents that understand the physical world. NVIDIA is adding post-training and optimization tools that help developers improve world models using less compute, plus simulation checkpoints for evaluating behavior in interactive driving and robotic manipulation scenarios before real-world testing. Cosmos Data Factory, made up of blueprints and core agent skills, helps curate, label, generate and evaluate training data. Porsche with Voxel51, ANYbotics with Encord, Zalando and CEVA Logistics with Sereact, and Siemens are using it.

Alpamayo 2 Edge is a compact reasoning model for real-time scene understanding and motion planning in vehicles. It is built on the Nemotron-based foundation of Cosmos 3 Edge and optimized for the NVIDIA DRIVE AGX Thor platform, with lower computing requirements.

In biology, BioNeMo RFD4-Proteina is an open protein design model from NVIDIA and the Baker Lab at the University of Washington's Institute for Protein Design. According to NVIDIA, it delivers leading performance across more than 10 protein design tasks and generates designs about 13x faster than RFdiffusion3 in a benchmark of 200 amino acid proteins.

The Ising family targets quantum development, from real-time hardware calibration to error correction. IQM Quantum Computers uses it to calibrate its processors, SEEQC combines it with NVQLink and its Digital Link for real-time error correction, and AWS is integrating Ising into AWS Parallel Computing Service so researchers can build on it before deployment.

Availability and what it means

Nemotron 3.5 Super VL is available on Hugging Face, ModelScope, OpenRouter and build.nvidia.com as an NVIDIA NIM microservice. Nemotron 3.5 Bolt E4B 4GB was expected on Friday, Oct. 23. Nemotron Retriever, Kumo, RFD4-Proteina and Ising are available through Hugging Face, build.nvidia.com or GitHub, depending on the model. Cosmos is openly available on Hugging Face and GitHub, and Alpamayo 2 Edge was slated for GitHub and Hugging Face later in October.

For European enterprises, the pitch is control: open weights and datasets that can be adapted to local data and run on their own infrastructure. The range of named partners suggests NVIDIA wants these models used across software, finance, manufacturing, logistics and research rather than in a single sector.