NVIDIA Launches Nemotron 3 Nano Omni Multimodal Model with 9x Throughput Boost

Gate News message, April 28 — NVIDIA has released Nemotron 3 Nano Omni, an open-source multimodal model featuring a 30B-A3B mixture-of-experts (MoE) architecture with 256K context window support. The model unifies processing of video, audio, image, and text inputs in a single framework.

Compared to comparable open-source multimodal models, Nemotron 3 Nano Omni achieves a 9x throughput improvement, significantly reducing inference costs and enhancing scalability. The model is now available on Hugging Face, OpenRouter, and NVIDIA NIM, and has been adopted by enterprises including Aible, Applied Scientific Intelligence, and H Company.