NVIDIA Releases Nemotron 3 Ultra for Enhanced Agentic AI

NVIDIA has launched Nemotron 3 Ultra, a 550B-parameter Mixture-of-Experts model with 55B active parameters, specifically designed for orchestrating complex, long-running agent workflows. It features architectural innovations like hybrid Mamba-Transformer layers for efficient long-context handling and NVFP4 quantization for cross-architecture GPU deployment, offering up to 5x higher throughput. The model also uses Multi-Teacher On-Policy Distillation (MOPD) for continuous improvement by learning from over ten domain-specific teacher models. NVIDIA is releasing Nemotron 3 Ultra as an open model under OpenMDW-1.1, with open weights, data, and recipes.