NVIDIA launches Nemotron 3.5 Lightning aimed at accelerating AI agents.

NVIDIA will expand the Nemotron 3 model series with the launch of the Nemotron 3.5 Lightning. Compared to other models in its class, this model can increase output speed by up to 4 times and improve task completion speed for agents by 30%. NVIDIA describes the Nemotron 3.5 Lightning as the most efficient model for long-duration agent AI workloads in its class. The Nemotron 3.5 Lightning is a mixed expert model with 30 billion parameters, designed to help develop smarter and more efficient agent applications. The Nemotron 3.5 Lightning is now available as NVIDIA NIM microservices on Hugging Face, ModelScope, OpenRouter, and build.nvidia.com.