Key Moments
- Nvidia introduced a new generation of its open-source “Nemotron” large-language models, touting higher speed, lower operating costs and improved capabilities over earlier versions.
- The smallest model, Nemotron 3 Nano, was released on Monday, with two larger variants scheduled for launch in the first half of 2026.
- The rollout comes as open-source models from Chinese firms such as DeepSeek, Moonshot AI and Alibaba Group Holdings gain traction, while many U.S. states and agencies restrict their use.
Overview of Nvidia’s New Open-Source AI Suite
SAN FRANCISCO, Dec 15 (Reuters) – Nvidia on Monday introduced a new family of open-source artificial intelligence models that it says will be faster, less expensive to operate and more capable than its earlier lineup, at a time when open-source AI offerings from Chinese developers are spreading rapidly.
Best known for its chips that power training of closed-source models from companies such as OpenAI, Nvidia also develops a broad range of its own AI models. These span use cases including physics simulations and autonomous driving systems and are made available as open-source software for researchers and commercial users. Firms such as Palantir Technologies have integrated Nvidia’s models into their own products.
$NVDA going open source
NVIDIA Debuts Nemotron 3 Family of Open Models
The model family includes:
Nemotron 3 Nano, a small 30-billion-parameter model that activates up to 3 billion parameters at a time for targeted, highly efficient tasks.Nemotron 3 Super, a… pic.twitter.com/tbyHIOdI2S
— Lon Juricic (@lonjuricic) December 15, 2025
Nemotron 3: New Generation of Large-Language Models
The company unveiled the third generation of its “Nemotron” large-language models, which are designed for applications such as writing, software development and other generative AI tasks. The smallest version in the lineup, Nemotron 3 Nano, was released Monday. Nvidia said two larger Nemotron 3 models are planned for release in the first half of 2026.
Nvidia, which described itself as the world’s most valuable publicly traded company, said Nemotron 3 Nano delivers higher efficiency than the prior generation. According to the company, this translates into lower operating costs and better performance on extended, multi-step tasks.
| Model Family | Version | Release Timing | Key Characteristics (as stated) |
|---|---|---|---|
| Nemotron 3 | Nemotron 3 Nano | Released Monday | More efficient than predecessor; cheaper to run; improved handling of long, multi-step tasks |
| Nemotron 3 | Larger variants (2 models) | First half of 2026 | Details not specified; part of new Nemotron 3 family |
Competitive Landscape: U.S. and Chinese Open-Source Models
Nvidia’s latest launch arrives as open-source AI models from Chinese technology companies, including DeepSeek, Moonshot AI and Alibaba Group Holdings, become increasingly prevalent in the technology sector. Companies such as Airbnb have disclosed that they are using Alibaba’s Qwen open-source model.
Meanwhile, CNBC and Bloomberg have reported that Meta Platforms is weighing a move toward closed-source models. That potential shift would leave Nvidia positioned as one of the more visible U.S.-based providers of open-source AI offerings.
The growth of Chinese open-source models has also intersected with regulatory concerns. Many U.S. states and government organizations have prohibited the use of Chinese AI models, citing security risks.
Nvidia’s Strategy: Transparency and Customization
Kari Briski, vice president of generative AI software for enterprise at Nvidia, said the company’s goal is to deliver a “model that people can depend on”, emphasizing that Nvidia is also making its training data and related tools openly available. This approach is intended to allow both government and corporate users to evaluate the models for security and tailor them to their own requirements.
TradingPedia.com is a financial media specialized in providing daily news and education covering Forex, equities and commodities. Our academies for traders cover 
