NVIDIA debuts Nemotron 3.5 Lightning as first open model
NVIDIA said it used distillation to give Nemotron 3.5 Lightning capabilities comparable to its larger Nemotron models.
NVIDIA is launching Nemotron 3.5 Lightning, a lightweight open model that the company says can run on a single PC GPU. Companies can download, use and modify the model for free, without seeking permission from NVIDIA.
Nemotron 3.5 Lightning targets specialized workloads in always-on, multi-agent AI systems. The chipmaker says the model offers up to 4x faster output and 30% faster agentic task completion than competing models in its class.
Its open and customizable design allows organizations to post-train the model using proprietary data, tools and workflows and deploy it across PCs, edge devices, workstations, data centers and the cloud.
The release follows NVIDIA CEO Jensen Huang’s growing advocacy for open AI models. He has argued that open models encourage competition, reduce prices and strengthen AI sovereignty, while also benefitting NVIDIA by increasing the amount of AI software that runs on GPUs.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
NVIDIA is also launching NeMo Switchyard, a tool that automatically determines which AI model is the most appropriate and cost-effective for a given task.
NeMo Switchyard adds intelligent routing to agent applications, directing requests to the most capable and efficient model for each task without requiring developers to rewrite their applications. NVIDIA said internal testing showed frontier-level accuracy while reducing task-completion costs to nearly one-third of running Opus 4.8 alone.