NVIDIA expands Nemotron open models for agentic AI

NVIDIA has announced Nemotron 3.5 Lightning and NeMo Switchyard to improve open-model agentic AI performance and routing.

• Nemotron 3.5 Lightning is a 30-billion-parameter mixture-of-experts model for high-volume, specialized agentic tasks; NVIDIA says it delivers up to 4x faster output speed and 30% faster agentic task completion versus comparable models.
• NeMo Switchyard is an open source routing library for AI agents that automatically sends prompts to the most suitable model; NVIDIA says internal benchmarks show it can reduce task completion cost to nearly one-third of Opus 4.8 alone.
• The release says Lightning is available via Hugging Face, ModelScope, OpenRouter, build.nvidia.com and NVIDIA NIM, while Switchyard is available on GitHub; NVIDIA also highlights ecosystem partners including CrowdStrike, Harvey, CodeRabbit, Lila Sciences, Fastino Labs, Boomi, Cadence, Classmethod, Cognition, Kong, LangChain, LiteLLM, Nous Research, Ramp and Siemens.