nemotron-3.5-lightning

NVIDIA Nemotron 3.5 Lightning

NVIDIA Nemotron 3.5 Lightning is a customizable open model built to power always-on agents. Distilled from NVIDIA's frontier Nemotron 3 Ultra model, its hybrid Mixture-of-Experts architecture combines 30B total parameters with 3B active parameters to bring capable agentic AI to local systems, the edge, datacenters, and the cloud.

Highlights

• Built for always-on agents: trained for popular agent harnesses and designed for long-running, multi-turn workflows that gather context, reason, and act across many steps.
• Fast specialized task completion: supports multi-token prediction and is designed for fast token generation and efficient token rollout, helping always-on agents complete more steps and finish specialized tasks faster.
• Open and customizable: organizations can own the model, post-train it for specialized workflows, and retain control over model behavior, data handling, and deployment.
• Long-context workflows: supports up to a 1M-token context window for long-running, multi-turn workflows.
• Flexible deployment: designed to run locally on DGX Spark, DGX Station, RTX PRO, and RTX; at the edge on NVIDIA Jetson; and in datacenter and cloud environments.