NVIDIA Releases Nemotron 3.5 Lightning for Fast, Accurate Execution in Long-Running Agents

NVIDIA has released Nemotron 3.5 Lightning, an open 30B mixture-of-experts model with 3B active parameters, designed for high-volume execution tasks in long-running AI agents, such as tool calls and result validation. The model aims to reduce latency and cost while maintaining accuracy.