Nvidia’s best model is now live
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Nvidia has released Nemotron 3 Ultra, a 550-billion-parameter open-weight mixture-of-experts model with 55 billion active parameters, on Hugging Face, ModelScope, OpenRouter, and build.nvidia.com. Built on the Mamba 2 architecture, it supports context windows up to 1 million tokens and is optimized for long-running agentic workflows involving planning, tool use, and complex reasoning. Nvidia claims it is the fastest U.S. open-weight model and up to 30% cheaper than comparable models. However, benchmarks show it trails leading Chinese open-weight models by a few points and scores significantly below GPT-5.5 on the GDPVal real-world task benchmark. The model was trained on 14.8 trillion tokens, supports 12 natural languages and 43 programming languages, and is released under the OpenMDW-1.1 license with weights, datasets, and training recipes available.
