NVIDIA Expands Open Model Families to Power the Next Wave of AI

Advanced and SupercomputingAI & Machine LearningSocial Media & Digital Platforms

NVIDIA's latest advancements in AI and accelerated computing are poised to transform industries and revolutionize the way we live and work.

TL;DR
• NVIDIA expands open model families to power agentic, physical, and healthcare AI.
• New models accelerate innovation across digital and physical industries.
• NVIDIA's AI and accelerated computing technologies have numerous applications and benefits.

Summary
NVIDIA expands its open model families to accelerate agentic, physical, and healthcare AI, launching new multimodal, reasoning, and biomedical models.
• Nemotron 3 Ultra, Omni, and VoiceChat deliver multimodal intelligence with 5x throughput efficiency on Blackwell.
• Cosmos 3, Isaac GR00T N1.7, and Alpamayo 1.5 enable robots and autonomous vehicles to perceive, reason, and act in complex environments.
• Proteina-Complexa and nvQSP accelerate protein drug discovery and simulation, adding 30 million protein complex predictions.

The Nemotron family supports natural conversations, advanced visual reasoning, and real‑time voice interaction, with safety models that detect unsafe content and a retrieval pipeline that improves relevance. Companies such as CodeRabbit, CrowdStrike, and ServiceNow deploy these models for enterprise‑scale AI assistants, while Edison Scientific uses Nemotron in its Kosmos autonomous AI scientist.

Physical AI models integrate vision, language, and action, allowing humanoid robots and autonomous vehicles to navigate and adapt. NVIDIA Cosmos 3 unifies synthetic world generation with action simulation, and Isaac GR00T N1.7 is commercially viable for real‑world humanoid deployment. HCLTech, Milestone Systems, and Toyota Research Institute use these models to train and analyze physical AI systems.

Healthcare AI benefits from BioNeMo’s open platform, where Proteina-Complexa designs protein binders and nvQSP simulates treatment scenarios 77x faster than CPU baselines. Collaborations with EMBL, DeepMind, and Seoul National University expanded the AlphaFold database with 1.7 million high‑confidence predictions, speeding drug target discovery.

Open models are available on GitHub, Hugging Face, and NVIDIA NIM microservices, enabling secure, scalable deployment from edge to cloud. NVIDIA’s strategy emphasizes transparency, safety, and continuous improvement, ensuring that AI systems augment human capabilities while mitigating long‑term risks.

As NVIDIA continues to push the boundaries of AI and accelerated computing, we must consider the potential implications and opportunities for innovation and growth.