Nvidia launches Nemotron 3 Nano Omni multimodal model for agentic AI

Nvidia Corp. on Tuesday launched Nemotron 3 Nano Omni, an open multimodal reasoning model that unifies text, vision, and speech processing in a single architecture. The model weighs about 30 billion parameters and uses a 30B-A3B hybrid mixture-of-experts architecture, combining vision and audio encoders to eliminate separate perception modules. Nvidia said the design delivers up to nine times faster throughput than other open omni models on the market, while remaining small enough to run on higher-end consumer hardware and enterprise cloud deployments. Nvidia positioned the model as the "brains" of faster, smarter agentic AI applications. Gautier Cloix, chief executive of H Company, said in a statement provided The new model is designed to run alongside other Nvidia Nemotron open models, such as Nemotron 3 Super for high-frequency execution or Super for complex planning. It is available on Hugging Face, OpenRouter, and build.nvidia.com as an Nvidia NIM microservice. Nvidia said the Nemotron family, including Ultra, Super, and Nano variants, has seen over 50 million downloads in the past year.