AI agent systems today juggle separate models for vision, speech and language — losing time and context as they pass data from one model to the other. Unveiled today, NVIDIA Nemotron 3 Nano Omni is an open multimodal model that brings these capabilities together into one system, enabling agents to d
