NVIDIA Nemotron 3 Nano Omni: Open 30B MoE Model With 3B Active for Edge AI Agents

NVIDIA·medium signal

NVIDIA released Nemotron 3 Nano Omni, an open multimodal model with 30B total parameters but only 3B active per forward pass via mixture-of-experts, designed for edge deployment in agentic workflows. The model tops six benchmarks across document intelligence, video understanding, and audio comprehension, delivering 9x higher throughput than comparable open multimodal models with accuracy jumping from 11.1 to 47.4 on the OSWorld GUI agent benchmark.