NVIDIA said it is expanding its open model portfolio across language, robotics, biology and quantum computing, positioning the release as infrastructure for developers and enterprises building applications tailored to local languages, industries and controlled systems. The company said the package is intended to support what it calls sovereign intelligence across Europe, while extending its developer stack into structured data, autonomous driving and scientific research.
Evidence and context
In material published on Wed, 21 Oct 2026, NVIDIA said the portfolio includes Nemotron 3.5 Super VL and Nemotron 3.5 Bolt; Nemotron Retriever models for embedding, reranking and document parsing; the Kumo family for structured business data; Cosmos 3.5 world models; Alpamayo 2 Edge for autonomous vehicles; BioNeMo for healthcare and scientific discovery; and Ising for quantum computing. NVIDIA said these open models can be customized with local data and deployed on infrastructure that organizations control.
For language and agentic workflows, NVIDIA said Nemotron includes open language models, datasets and recipes, with support for text and images. The company said the models are built with data, research and expertise from the NVIDIA Nemotron Coalition, including European partners Black Forest Labs, H-Company and Mistral. NVIDIA said 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. It also listed a wider set of post-training partners, including AfterQuery, AgileRL, AI21, Alice, Applied Compute, Baseten, CoreWeave, Distil Labs, EON, Fastino, Fireworks, Google Cloud, Mercor, Multiverse Computing, Nebius, Handshake AI, Plurai, Prime Intellect, River.AI, Scale AI, Trajectory and TNG.
For structured enterprise data, NVIDIA said Kumo models and tools are designed to forecast outcomes, score fraud risk and recommend actions from relational and tabular data. The company said Kumo Structured covers relational and tabular prediction, forecasting and anomaly detection, while Kumo Engine is an open library stack for developing and deploying structured data foundation models. NVIDIA said Cognite, Coinbase, Databricks, DoorDash, Revolut and Tubi are among the first to work with Kumo models, and that Snowflake customers are already using them in advertising, energy, financial services and retail.
In robotics and simulation, NVIDIA said Cosmos is an open world-model family for robots, autonomous vehicles and vision agents. The company said new post-training and optimization tools are intended to improve world models using less compute, while simulation checkpoints are meant to evaluate behavior before real-world testing. NVIDIA also introduced Cosmos Data Factory, which it said includes blueprints and agent skills for curating, labeling, generating and evaluating training data for driving scenarios, product defects, image augmentation and event generation. NVIDIA said Porsche with Voxel51, ANYbotics with Encord, Zalando and CEVA Logistics with Sereact, and Siemens are using Cosmos Data Factory.
NVIDIA said Alpamayo 2 Edge is a compact reasoning model for real-time scene understanding and motion planning in autonomous vehicles, built on Cosmos 3 Edge’s Nemotron-based foundation and optimized for the NVIDIA DRIVE AGX Thor platform. In biology, NVIDIA said BioNeMo RFD4-Proteina was developed with the Baker Lab at the University of Washington’s Institute for Protein Design and can generate protein sequences and 3D structures. NVIDIA said the model delivers performance across more than 10 protein design tasks and generates designs about 13x faster than RFdiffusion3 in a benchmark of 200 amino acid proteins.
For quantum computing, NVIDIA said the Ising family is intended to accelerate hardware calibration and error correction across qubit modalities. The company said IQM Quantum Computers is using Ising to calibrate processors, that SEEQC is combining NVQLink and Ising with its Digital Link for low-latency error correction, and that AWS is integrating Ising into AWS Parallel Computing Service.
Availability varies by model. NVIDIA said Nemotron 3.5 Super VL is available on Hugging Face, ModelScope, OpenRouter and build.nvidia.com as an NVIDIA NIM microservice, along with NVIDIA cloud, post-training and inference partners. Nemotron 3.5 Bolt E4B 4GB is expected on Fri, 23 Oct 2026. NVIDIA said Nemotron Retriever, Kumo, Cosmos, BioNeMo RFD4-Proteina and Ising are available on combinations of Hugging Face, GitHub, build.nvidia.com and select partner platforms, while Alpamayo 2 Edge is due for download later this month.
Recent NVIDIA headlines on 15 Oct 2026 and 14 Oct 2026 also pointed to earlier Nemotron-family expansion, but those items are separate from this latest model portfolio announcement. The source material does not provide independent performance verification beyond NVIDIA’s own claims, and it does not quantify adoption in Europe beyond the company’s framing of the release.
Be the first to comment