NVIDIA has launched the Nemotron 3 family, featuring Nano, Super, and Ultra models built on a hybrid latent mixture-of-experts (MoE) architecture. The 30-billion-parameter Nano model, available now, promises up to four times higher token throughput than its predecessor and is optimized for low-cost inference. CEO Jensen Huang emphasized that "open innovation is the foundation of AI progress," as the models aim to enable transparent, efficient multi-agent systems across various industries.
NVIDIA has launched the Nemotron 3 family, featuring Nano, Super, and Ultra models built on a hybrid latent mixture-of-experts (MoE) architecture. The 30-billion-parameter Nano model, available now, promises up to four times higher token throughput than its predecessor and is optimized for low-cost inference. CEO Jensen Huang emphasized that "open innovation is the foundation of AI progress," as the models aim to enable transparent, efficient multi-agent systems across various industries.