Nvidia has launched Nemotron 3 Super, an open-weight AI model featuring 120 billion total parameters. The model utilizes a hybrid Mixture-of-Experts architecture that activates only 12 billion parameters during inference. Nvidia designed the system specifically for complex, agent-like AI applications. The company is releasing the full weights, datasets, and training methodologies to the developer community.
A regulatory filing reveals Nvidia plans to invest $26 billion into the development of open-weight AI models. This capital commitment signals a strategic shift from hardware infrastructure provider to a direct competitor of leading developers like OpenAI and Anthropic. The move positions Nvidia as a primary player in the AI model landscape.