On Monday, Nvidia unveiled a new family of open-source artificial intelligence models that it says will be faster, cheaper, and smarter than its previous offerings, at a time when open-source options from Chinese AI labs are proliferating. Nvidia is primarily known for providing the chips used by companies like OpenAI to train their closed-source models.
Article contents:
- Launch of the new Nemotron family
- Competition with Chinese models
- Open source vs. closed source
- Nvidia’s strategic shift
- Security and trust as a draw
- The future of open-source software
- Frequently asked questions
Launch of the new Nemotron family
Nvidia revealed the third generation of its “Nemotron” large language models aimed at writing, coding, and other tasks. The smallest model, named “Nemotron 3 Nano,” was released on Monday, with two larger versions coming in the first half of 2026. The company, which has become the world’s most valuable, said that “Nemotron 3 Nano” is more efficient than its predecessor—meaning cheaper operation—and will deliver better performance in long, multi-step tasks.
Competition with Chinese models
Nvidia is releasing the models as open-source offerings from Chinese tech companies such as DeepSeek, Moonshot AI, and Alibaba Group are widely used in the tech industry. Companies like Airbnb have disclosed using Alibaba’s open-source “Qwen” model. Many U.S. states and government bodies have banned the use of Chinese models due to security concerns, creating a market gap for secure alternatives.
Open source vs. closed source
Meanwhile, reports from CNBC and Bloomberg indicated that Meta is considering a shift toward closed-source models, leaving Nvidia as one of the most prominent U.S. providers of open-source offerings. This shift could reshape the landscape, as Meta was previously a champion of open source with its “Llama” models.
Security and trust as a draw
Cari Briskey, vice president of enterprise generative AI software at Nvidia, said the company aims to provide “a model that people can rely on,” also releasing training data and other tools openly so government and commercial users can test them for security and customize them. Briskey told Reuters: “That’s why we treat it like a library. And that’s why we’re committed to it from a software engineering perspective.”
Frequently asked questions
Q: What is an open-source model?
A: It is an artificial intelligence model whose source code and data are made publicly available for free, allowing developers to modify, improve, and use it in their own applications.
Q: Why does Nvidia care about launching software when it is a hardware company?
A: To ensure continued demand for its chips; the more models available and the easier they are to use, the more companies need to buy Nvidia processors to run them.
Q: What is the difference between Nemotron and Chinese models?
A: Nemotron offers a secure and reliable American alternative for companies and governments avoiding the use of Chinese technology for security and political reasons.
