Nemotron 3.5 Lightning is a 30B MoE built for the boring roles inside multi-agent systems.

NVIDIA claims up to 4x faster output and 30% faster agentic task completion versus comparable models, aimed explicitly at high-volume narrow work: code review, tool use, security monitoring, billing triage (NVIDIA). They also published Nemotron-RL-Agentic-Terminal-Pivot, an agentic RL dataset for coding-agent post-training. The model is fine. Switchyard, the routing library that dispatches each request to the cheapest model that can handle it, is the artifact that changes how you build.