NVIDIA Launches Open Model Portfolio to Advance Sovereign Intelligence Across Industries in Europe

Summary:NVIDIA has expanded its open model portfolio across language, robotics, biology, and quantum computing. The new releases include NVIDIA Nemotron 3.5 Super VL and Bolt models, Nemotron Retriever for embedding, reranking, and document parsing, NVIDIA Kumo for structured business data, NVIDIA Cosmos 3.5 world models, NVIDIA Alpamayo 2 Edge for autonomous vehicles, NVIDIA BioNeMo for healthcare and scientific discovery, and NVIDIA Ising for quantum computing. These models provide a foundation for de

---

NVIDIA Launches Open Model Portfolio

NVIDIA has announced the expansion of its open model portfolio across multiple domains, including language, robotics, biology, and quantum computing. This series of releases aims to provide developers, researchers, and enterprises with the tools to build tailored solutions for local languages and industries.

Overview of Key Releases

1. NVIDIA Nemotron 3.5 Super VL and Bolt

   • These models offer powerful language processing capabilities, supporting text and image inputs, and help agents orchestrate workflows, solve complex problems, and perform specialized tasks.

2. NVIDIA Nemotron Retriever

   • This series focuses on embedding, reranking, and document parsing, providing a foundation for building efficient information retrieval systems.

3. NVIDIA Kumo

   • Kumo is a set of open foundation models, libraries, and tools for structured business data. It helps enterprises accelerate AI development and reduce the cost and complexity of building machine learning pipelines by predicting outcomes, scoring fraud risk, and recommending actions.

4. NVIDIA Cosmos 3.5

   • Cosmos is a family of world models that help developers build robots, autonomous vehicles, and vision agents that understand and interact with the physical world. NVIDIA has also introduced new Cosmos post-training and optimization tools, as well as Cosmos Data Factory for curating, labeling, generating, and evaluating training data.

5. NVIDIA Alpamayo 2 Edge

   • This is a compact reasoning model that brings real-time scene understanding and motion planning directly into autonomous vehicles. It is built on NVIDIA Cosmos 3 Edge's Nemotron foundation and optimized for the NVIDIA DRIVE AGX Thor platform.

6. NVIDIA BioNeMo RFD4-Proteina

   • This is an open protein design model developed by NVIDIA and the Baker Lab at the University of Washington’s Institute for Protein Design. The model generates protein sequences and 3D structures to help researchers design proteins that bind to difficult targets, assemble into complex structures, and bind to small molecules.

7. NVIDIA Ising

   • The Ising family of open models provides a common, extensible foundation for quantum computing, covering aspects from real-time hardware calibration to error correction, and tailored to different qubit modalities and hardware needs.

Engineering Trade-offs and Practical Applications

• Nemotron Series: By supporting multimodal inputs and efficient inference, Nemotron models maintain high performance while reducing computational costs. This makes them suitable for a wide range of industry applications, including intelligent customer service, automated document processing, and data analysis.

• Kumo Models: Kumo simplifies structured data processing tasks through pre-trained models and specialized libraries. However, it may require additional customization when dealing with highly unstructured data.

• Cosmos Models: Cosmos provides powerful world model support for robotics and autonomous driving, but its training and optimization processes require significant computational resources.

• BioNeMo RFD4-Proteina: The model excels in protein design tasks, but its ability to generate complex protein structures still needs to be validated in the laboratory.

• Ising Models: Ising provides specialized support for quantum computing hardware calibration and error correction, but its application is currently limited to specific hardware platforms.

Developer Deployment Recommendations

• Multimodal Support: Developers can leverage Nemotron and Cosmos models to build more intelligent and interactive applications by utilizing their multimodal input capabilities.

• Structured Data Processing: Kumo models are suitable for applications that require processing large amounts of structured data, such as financial analysis, supply chain management, and market forecasting.

• Resource Management: Cosmos and Ising models require high computational resources. Developers should consider using NVIDIA's dedicated hardware and cloud services to optimize deployment.

• Cross-Domain Applications: These open models provide rich possibilities for cross-domain applications. Developers can combine the strengths of different models to create innovative solutions.

Conclusion

NVIDIA's open model portfolio offers new impetus for the widespread application and industry innovation of AI technology. By providing powerful tools and resources, NVIDIA is driving the deep integration and development of AI technology across various fields.

---

Source: NVIDIA AI News (2026-10-21)

— END —

Tags: #NVIDIA Follow #Open Models Follow #Multidomain AI Follow #Nemotron Follow #Kumo Follow #Cosmos Follow #Alpamayo Follow #BioNeMo Follow #Ising Follow