How to Run Open-Weight Nemotron 3 Models on a GPU Droplet

This article was original written by Andrew Dugan (Senior AI Technical Content Creator II) NVIDIA has announced the newest additions to their Nemotron family of models, Nemotron 3 . There are three separate models in the Nemotron 3 family that are being released, including Nano, Super, and Ultra, which have 30, 49, and 253 billion parameters respectively with up to 1M tokens in context length. Nano was released in December of 2025, and Super and Ultra are scheduled to be released later in 2026. They are being released on NVIDIA’s open model license , making them available for commercial use and modification and giving you ownership and complete control over generated outputs. Both the weights and training data are open and available on Hugging Face . This tutorial will discuss these models and how to deploy the currently available Nano on a DigitalOcean GPU Droplet . Key Takeaways NVIDIA has announced Nemotron 3, a new addition to their Nemotron model lineup. Nemotron 3 consists of thr