.svg){kind=spacer}
Details
- NVIDIA has expanded its Nemotron and Cosmos AI model families, unveiling new releases at SIGGRAPH aimed at advancing agentic and physical AI applications for enterprises.
- Major organizations such as CrowdStrike, Uber, and Zoom are integrating these models, leveraging them for uses ranging from AI-driven cybersecurity agents to improved autonomous vehicle analysis and smarter meeting assistants.
- The update introduces the Nemotron Nano 2 and Llama Nemotron Super 1.5, promising improved performance metrics such as faster token generation and lower computational costs, enabled by new hybrid architectures and configurable processing options.
- Cosmos Reason, a newly announced vision-language model, features 7 billion parameters and offers advanced physical and spatial reasoning ideal for robotics and autonomous systems, expanding on NVIDIA’s earlier 2025 launches.
- NVIDIA's latest models will be available as cloud-hosted microservices via Amazon Bedrock, Azure AI Foundry, and Google Vertex AI, alongside the company’s first open vision-language model training set featuring 3 million samples.
Impact
NVIDIA’s bold expansion into agentic and physical AI solidifies its leadership in an enterprise market projected to reach $450 billion by 2028. Enhanced reasoning capabilities and high-profile enterprise adoption set NVIDIA apart from rivals, accelerating practical deployment of autonomous AI agents across sectors including security, mobility, and communications.
