Nvidia Unveils New Open-source AI Models Amid Boom In Chinese Offerings

SanFrancisco: Nvidia on Monday introduced a new family of open-source AI models that it says will be faster, cheaper and smarter than its previous offerings, as open-source offerings from Chinese AI labs proliferate.

Nvidia is known primarily for supplying the chips that companies like OpenAI use to train closed-source models and charge money for them. But it also offers its own models for everything from physics simulations to self-driving vehicles to open-source software that can be used by researchers or other companies; Companies like Palantir Technologies are incorporating Nvidia’s model into their products.

Nvidia on Monday introduced the third generation of its “Nemotron” large-language models aimed at authoring, coding and other tasks. The smallest of the models, called Nemotron 3 Nano, was being released on Monday; Two larger versions would arrive in the first half of 2026.

Nvidia, which has become the world’s most valuable listed company, said the Nemotron 3 Nano is more efficient than its predecessor, meaning it will be cheaper to run and perform better in long multi-step tasks.

Nvidia is launching the models as open source offerings from Chinese tech firms such as DeepSeek, Moonshot AI and Alibaba Group Holdings begin to become widely used in the tech industry, and as companies such as Airbnb announce the use of Alibaba’s Qwen open source model.

At the same time, CNBC and Bloomberg reported that Meta Platforms was considering moving to closed source models, and Nvidia had become one of the leading open source providers in the US.

Many US states and government agencies have banned the use of Chinese models due to safety concerns.

Kari Briski, Nvidia’s vice president of enterprise generative AI software, said the company aims to provide “a model that people can trust” and is also openly publishing training data and other tools so government and business users can test and customize for security.

“So we treat it like a library,” Briski said in an interview with Reuters. “So we’re committed to this from a software engineering perspective.”