Nvidia developing Nemotron 4 AI model

Nvidia is developing a new AI model family, Nemotron 4, with the goal of rivaling top open-source models globally, according to Information (disambiguation). The largest model is expected to have at least 1 trillion parameters, and could be ready as early as late fall, though no release date has been set. Nvidia also unveiled Nemotron 3.5 Lightning, aimed at code review, tool use, security alert monitoring, and billing questions, and released NeMo Switchyard, an open-source model-routing library. These moves come as Nvidia expands its open-source AI push amid rising AI costs and competition from Chinese models. Nvidia last month formed a coalition for AI safety and cybersecurity and signed an open letter with Microsoft backing open-weight models.