Key insights
- NVIDIA launched Nemotron 3 Nano Omni, a multimodal AI model integrating vision, audio, and language. It boasts high throughput and handles various inputs, potentially enhancing AI agent capabilities. The open-source release with training data allows customization and deployment flexibility. Positive for NVDA and AI sector, suggesting continued innovation and market leadership.
Investing.com -- NVIDIA (NASDAQ:NVDA) introduced Nemotron 3 Nano Omni on Tuesday, an open multimodal model that unifies vision, audio and language capabilities into a single system for AI agents.
The model combines vision and audio encoders within a 30B-A3B hybrid mixture-of-experts architecture, eliminating the need for separate perception models. According to NVIDIA, this design enables AI systems to achieve up to 9x higher throughput than other open omni models with similar interactivity.
Nemotron 3 Nano Omni handles text, images, audio, video, documents, charts and graphical interfaces as inputs, producing text as output. The model features a 256K context window and uses Conv3D and EVS technologies.
The model topped six leaderboards for document intelligence, video and audio understanding, NVIDIA said.
AI and software companies adopting the model include Aible, Applied Scientific Intelligence, Eka Care, Foxconn, H Company, Palantir and Pyler. Dell Technologies, DocuSign, Infosys, K-Dense, Lila, Oracle and Zefr are evaluating it.
Gautier Cloix, CEO of H Company, said the model enables agents to rapidly interpret full HD screen recordings. In preliminary evaluations on the OSWorld benchmark, H Company’s computer usage agent powered by Nemotron 3 Nano Omni used a native input resolution of 1920×1080 pixels for visual reasoning.
The model is designed to work alongside other NVIDIA Nemotron models, such as Nemotron 3 Super and Ultra, or proprietary models from other providers. It supports agentic workflows including computer use, document intelligence and audio-video reasoning.
NVIDIA released Nemotron 3 Nano Omni with open weights, datasets and training techniques. Organizations can customize the model using tools like NVIDIA NeMo and deploy it in environments meeting regulatory or data localization requirements.
The Nemotron 3 family has seen over 50 million downloads in the past year, according to NVIDIA.
Nemotron 3 Nano Omni is available Tuesday on Hugging Face, OpenRouter and build.nvidia.com as an NVIDIA NIM microservice, as well as through NVIDIA Cloud Partners, inference platforms and cloud service providers.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.