NVIDIA has launched a suite of new open models, data, and tools designed to accelerate AI innovation across industries. The company introduced a range of specialized AI solutions including the NVIDIA Nemotron family for agentic AI, the NVIDIA Cosmos platform for physical AI, the new NVIDIA Alpamayo family for autonomous vehicles, NVIDIA Isaac GR00T for robotics, and NVIDIA Clara for biomedical research. These resources are part of NVIDIA’s broader effort to expand the open model ecosystem and empower organizations to build real-world AI systems. The company is releasing one of the world’s largest collections of open multimodal data, including 10 trillion language training tokens, 500,000 robotics trajectories, 455,000 protein structures, and 100 terabytes of vehicle sensor data. This unprecedented scale of diverse open resources aims to drive progress in language models, robotics, scientific discovery, and autonomous systems. Leading technology firms including Bosch, CodeRabbit, CrowdStrike, Cohesity, Fortinet, Franka Robotics, Humanoid, Palantir, Salesforce, ServiceNow, Hitachi, and Uber are already adopting NVIDIA’s open technologies to advance their AI initiatives. NVIDIA Nemotron is being expanded with new models focused on speech, multimodal retrieval-augmented generation (RAG), and safety. Bosch is using Nemotron Speech to enable natural voice interactions in vehicles. ServiceNow is training its Apriel model family on open datasets, including Nemotron, to achieve cost-efficient multimodal performance. Cadence and IBM are testing Nemotron RAG models to enhance search and reasoning in complex technical documentation. CrowdStrike, Cohesity, and Fortinet are integrating Nemotron Safety models to improve the reliability and trustworthiness of their AI systems. Palantir is incorporating Nemotron into its Ontology framework to create a unified stack for specialized AI agents. CodeRabbit is leveraging Nemotron to scale AI-powered code reviews, increasing speed and accuracy while reducing costs. NVIDIA is also releasing open-source training code, datasets, and blueprints, such as the dataset and training code for the Llama Embed Nemotron 8B model, which ranks on the MMTEB leaderboard. The updated LLM Router helps developers route AI requests to the most suitable model, and a new dataset supports the development of the Nemotron Speech ASR model. For physical AI and robotics, NVIDIA is releasing the NVIDIA Cosmos open world foundation models, which enable humanlike reasoning and world generation to speed up the development and validation of physical AI systems. The company is also providing open models and blueprints for various physical embodiments built on Cosmos. Salesforce, Milestone, Hitachi, Uber, VAST Data, and Encord are using Cosmos Reason for AI agents focused on traffic and workplace productivity. Franka Robotics, Humanoid, and NEURA Robotics are using Isaac GR00T to simulate, train, and validate robot behaviors before deploying them in real-world settings. In the autonomous vehicle space, NVIDIA Alpamayo introduces a new family of open models, simulation tools, and large datasets to advance reasoning-based autonomous driving. This includes over 1,700 hours of driving data collected across diverse geographies and challenging conditions, capturing rare and complex edge cases critical for improving AI reasoning. For healthcare and life sciences, NVIDIA Clara introduces new AI models that accelerate drug discovery and treatment development. These models help researchers design safer, more effective therapies. The release also includes a dataset of 455,000 synthetic protein structures to support AI-driven scientific research. All NVIDIA open models, data, and frameworks are now available on GitHub and Hugging Face, as well as through cloud platforms and AI infrastructure providers. Developers can access supporting resources via build.nvidia.com. Many models are also available as NVIDIA NIM microservices for secure, scalable deployment across NVIDIA-accelerated environments—from edge devices to the cloud. More details can be found by watching NVIDIA Live at CES.