NVIDIA announces a large amount of AI-related software and hardware, including the open source AI model 'Alpamayo' and the AI chip 'Rubin,' which accelerate the development of self-driving cars

NVIDIA unveils new open models and data tools to advance AI across industries.

NVIDIA Unveils New Open Models, Data and Tools to Advance AI Across Every Industry | NVIDIA Blog
https://blogs.nvidia.com/blog/open-models-data-tools-accelerate-ai/

The 'Alpamayo' family is an open-source AI model and tool announced by NVIDIA to accelerate the development of safe, inference-based autonomous vehicles. The Alpamayo family employs a Vision, Language, Action (VLA) model based on thought chains and inference, enabling human-like thinking in autonomous vehicle decision-making. The Alpamayo family also includes simulation tools and datasets that can be used in autonomous vehicle development, which NVIDIA claims 'can also contribute to accelerating the development of autonomous vehicles.'

NVIDIA Announces Alpamayo Family of Open-Source AI Models and Tools to Accelerate Safe, Reasoning-Based Autonomous Vehicle Development | NVIDIA Newsroom
https://nvidianews.nvidia.com/news/alpamayo-autonomous-vehicle-development

As part of the Alpamayo family, NVIDIA has released the VLA model 'Alpamayo 1,' the fully open source end-to-end simulation framework 'AlpaSim' for high-fidelity autonomous vehicle development, and the diverse large-scale open datasets for autonomous vehicles 'Physical AI Open Datasets,' which can be accessed below.

nvidia/Alpamayo-R1-10B · Hugging Face
https://huggingface.co/nvidia/Alpamayo-R1-10B

GitHub - NVlabs/alpasim
https://github.com/NVlabs/alpasim

nvidia/PhysicalAI-Autonomous-Vehicles · Datasets at Hugging Face
https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles

Additionally, NVIDIA has announced that it plans to introduce its autonomous driving software, NVIDIA DRIVE AV, with enhanced Level 2 point-to-point driver assistance capabilities, in the United States by the end of 2026. NVIDIA's long-time partner, Mercedes-Benz, will be installing NVIDIA DRIVE AV in its new CLA.

NVIDIA DRIVE AV Software Debuts in All-New Mercedes-Benz CLA | NVIDIA Blog
https://blogs.nvidia.com/blog/drive-av-software-mercedes-benz-cla/

NVIDIA has announced open models to enable developers to avoid resource-intensive pre-training and focus on developing the next generation of AI-driven robots and autonomous machines. The three models announced this time are ' NVIDIA Cosmos Transfer 2.5 ' and ' NVIDIA Cosmos Predict 2.5 ,' open and fully customizable world models that enable physics-based synthetic data generation and robot policy evaluation in physics AI simulations, 'NVIDIA Cosmos Reason 2 ,' an open inference visual language model that enables intelligent machines to see, understand, and act in the physical world like humans, and ' NVIDIA Isaac GR00T N1.6 ,' an open inference visual language action model designed specifically for humanoids.

NVIDIA Releases New Physical AI Models as Global Partners Unveil Next-Generation Robots | NVIDIA Newsroom
https://nvidianews.nvidia.com/news/nvidia-releases-new-physical-ai-models-as-global-partners-unveil-next-generation-robots

NVIDIA also announced its next-generation AI chip, Rubin. Rubin is equipped with six chips: the NVIDIA Vera CPU , NVIDIA Rubin GPU, NVIDIA NVLink 6 Switch , NVIDIA ConnectX-9 SuperNIC ,NVIDIA BlueField-4 DPU , and NVIDIA Spectrum-6 Ethernet Switch . This significantly reduces AI training time and inference token costs. Compared to NVIDIA Blackwell, Rubin reduces inference token costs by up to 10x and requires four times fewer GPUs to train MoE models .

Infrastructure for Scalable AI Reasoning | NVIDIA Rubin Platform
https://www.nvidia.com/en-us/data-center/technologies/rubin/

NVIDIA Kicks Off the Next Generation of AI With Rubin — Six New Chips, One Incredible AI Supercomputer | NVIDIA Newsroom
https://nvidianews.nvidia.com/news/rubin-platform-ai-supercomputer

NVIDIA DGX SuperPOD Sets the Stage for Rubin-Based Systems | NVIDIA Blog
https://blogs.nvidia.com/blog/dgx-superpod-rubin/

NVIDIA also announced 'NVIDIA Clara,' a suite of AI models for healthcare and life sciences, including 'La-Proteina,' which enables the design of large proteins with atomic precision for research and drug candidate development, 'ReaSyn v2,' which ensures the synthesis of AI-designed drugs is practical by incorporating manufacturing blueprints into the discovery process, 'KERMT,' which predicts how potential drugs will interact with the human body, and 'RNAPro,' which predicts the complex 3D shapes of RNA molecules.

NVIDIA also announced a model based on the open AI model Nemotron 3.

Nemotron Speech is an open model that includes a new ASR model that enables real-time, low-latency speech recognition for live captioning and voice AI applications, and has been shown to deliver performance 10x faster than other models in its class.

Nemotron RAG , a multimodal search augmentation generation (RAG) , consists of novel embeddings and re-ranking vision language models (VLMs) that provide high-accuracy multilingual and multimodal data insights to enhance document search and information retrieval.

Nemotron Safety, which enhances the safety and reliability of AI applications, includes Llama Nemotron Content Safety with expanded language support and Nemotron PII , which can detect sensitive data with high accuracy.

NVIDIA announces the 'Nemotron 3' series, a compact and high-performance open AI model - GIGAZINE

NVIDIA announced its latest open model at CES 2026, one of the world's largest consumer electronics trade shows, and you can check out the presentation in the video below.

NVIDIA Live with CEO Jensen Huang - YouTube