Nvidia launches Nemotron 3.5 Lightning as US open model momentum picks up

Nvidia launched Nemotron 3.5 Lightning, a lightweight open model, and NeMo Switchyard, a routing library for AI agents, in a move that adds another open model to the mix for local compute.

In a post, Nvidia outlined Nemotron 3.5 Lightning its highest efficiency model designed for running agentic AI systems. Nvidia Nemotron model family filled the void for US open models, which are lagging Chinese alternative.

• Palantir, Nvidia bring Nemotron open models to sovereign AI
• Nvidia Nemotron: Much needed open-source model champion in US

The Nvidia announcement follows Meta's open source move with its Muse Glimmer rollout.

Key points about Nemotron 3.5 Lightning include:

• The model is 30-billion parameters with a mixture-of-experts architecture.
• Nemotron 3.5 Lightning is customizable and built for high-volume tasks and always-on agents.
• The model has up to 4x faster output speed and can be post-trained for specialized tasks.
• According to Nvidia, Nemotron 3.5 Lightning is part of an ensemble of Nvidia Nemotron models. Palantir recently noted that Nemotron models can outperform frontier models when armed with the right data, ontology and process knowhow.

To ride along with Nemotron 3.5 Lightning, Nvidia launched NeMo Switchyard, an open source library for routing AI agents.

Nvidia's Nemotron models with NeMo Switchyard are designed for efficiency. They can also be used in local deployments.

Partners using NeMo Switchyard include Boomi, Cognition, Kong, LangChain and Siemens among others.

Open models storyline

• Enterprises need better AI value metrics and vendors need better pricing models
• Enterprise AI grows up, vendors are noticing
• Anthropic: Opposed to open model ban, wants safety testing for all models
• Get ready for US open LLMs and just in time
• Rightsizing open models may cut your AI inference spend
• Moonshot AI launches Kimi K3
• DeepSeek's real impact happening now
• Here’s what we learned about AI projects from enterprise buyers so far
• AI inference costs are going to be a big concern: What's the fix?
• Why enterprise AI leaders need to bank on open-source LLMs