NewsBite analysis · AI, Startups, Tools
MarkTechPost
A 30B Open MoE with 3B Active Parameters, and NeMo Switchyard Model Router
NVIDIA has released the Nemotron 3.5 Lightning, a highly customizable 30B mixture-of-experts model designed for agentic tasks, alongside the NeMo Switchyard routing library. This innovative combination aims to enhance efficiency by optimizing how AI agents use resources. The lightweight model boasts 3B active parameters and is tailored for industries such as cybersecurity and healthcare, providing up to 4x faster outputs compared to competitors.
ExperimentConfidence: HighImmediate horizon
Published 12 Aug 2026, 08:40 UTC3 min read
Visit original article ↗Lead image — MarkTechPost
Summary
TL;DR
- Nemotron 3.5 Lightning is a 30B mixture-of-experts model with 3B active parameters.
- The model achieves up to 4x faster output speeds than similar-sized models.
- Lightning targets execution efficiency, focusing on high-volume tasks for AI agents.
- NeMo Switchyard is an open-source routing library optimizing AI agent workflows.
- Numerous companies across various sectors are customizing the Lightning model for their needs.
Interesting facts
- Nemotron is built on a hybrid Mamba-2 + MoE + Attention architecture.
- The model has a remarkable 1M-token context window.
- It allows solo developers and seed-stage startups to leverage state-of-the-art AI capabilities.