OFICIAL NVIDIA Newsroom

NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI

What happened
Based on NVIDIA Newsroom · Aug 11, 2026

NVIDIA introduces Nemotron 3.5 Lightning, a 30-billion-parameter model optimized for agentic AI tasks, and NeMo Switchyard, an open-source routing library to direct tasks to the most suitable model.

🚩 Report a correction ↗ Open source

Video

Video available

Key points
·
As AI shifts from chatbots to autonomous agents, open models are serving market demands for full control over where AI runs and how it’s deployed and evolves.
·
Today, NVIDIA is expanding its Nemotron 3 model family with Nemotron 3.5 Lightning, the highest-efficiency model in its class for long-running agentic AI workloads.
·
This release follows Nemotron 3 Nano and reflects NVIDIA’s commitment to continually improving open models for greater accuracy and speed.
·
Built for specialized tasks within larger multi-agent systems, Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts model, helps create smarter and more efficient agentic applications.
Key numbers
·
5 Lightning, a 30-billion-parameter mixture-of-experts model designed for long-running agentic AI workloads.
·
The model is optimized for specialized tasks within multi-agent systems, offering up to 4x faster output speed and 30% faster task completion compared to similar models.
·
5 Lightning, a 30-billion-parameter model optimized for agentic AI tasks, and NeMo Switchyard, an open-source routing library to direct tasks to the most suitable model.

NVIDIA has expanded its Nemotron 3 model family with Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts model designed for long-running agentic AI workloads. The model is optimized for specialized tasks within multi-agent systems, offering up to 4x faster output speed and 30% faster task completion compared to similar models. It supports local deployment on NVIDIA RTX PCs, DGX systems, and Jetson devices, as well as cloud and data center environments.

NeMo Switchyard, an open-source library, enables intelligent routing of AI agent requests to the most capable model based on task requirements. It allows enterprises to use a mix of open, proprietary, and NVIDIA models without rewriting applications. The library supports customization of routing algorithms to prioritize quality, latency, or cost, improving token efficiency in model ensembles.

Nemotron 3.5 Lightning is fully customizable and can be post-trained on domain-specific data using NVIDIA NeMo to enhance accuracy for specialized tasks. Early adopters, including CrowdStrike, Harvey with Trajectory, and CodeRabbit with Baseten, are using the model to improve performance in cybersecurity, legal services, and code review, respectively.

The Nemotron 3.5 Lightning model and NeMo Switchyard are available on Hugging Face, ModelScope, OpenRouter, and NVIDIA’s build.nvidia.com as an NIM microservice, with broader ecosystem support through cloud partners and inference platforms. NeMo Switchyard is accessible on GitHub, with integration planned for partner platforms.

Original source → Deals on Clipraptor.com →