NVIDIA Debuts Nemotron 3 Family of Open Models for Agentic AI

Summary

At CES 2026, NVIDIA announced the Nemotron 3 family of open models in three sizes — Nano, Super, and Ultra — designed specifically for building agentic AI applications. The family uses a breakthrough hybrid latent mixture-of-experts (MoE) architecture that NVIDIA describes as the most efficient open model family with leading accuracy for agentic workloads. Nemotron 3 Nano (30B parameters, up to 3B active) is available immediately via Hugging Face and various inference providers. It delivers 4x