NVIDIA launches its first open-source AI model

Key Points

1. NVIDIA released the Nemotron 3.5 Lightning open-source model, which the company claims is lightweight and can run on a single GPU of a laptop or desktop computer.
2. This is the first open-source model launched by the chip manufacturer since CEO Jensen Huang publicly supported the open-source approach last month.
3. In July, Huang stated in an interview: "Free and open artificial intelligence will benefit hardware and the chip industry."

On March 16, 2026, at the NVIDIA GTC conference in San Jose, California, CEO Jensen Huang holds the Spectrum-X Ethernet photonic chip.

At the end of July, NVIDIA CEO Jensen Huang made his first post on the X platform, publicly voicing support for open-source AI models and deeply engaging in the industry-wide debate.

Less than three weeks later, NVIDIA officially released the Nemotron 3.5 Lightning. The company introduced this model as lightweight, capable of running on a single graphics processor of a personal computer. This is NVIDIA's first open-source model since Huang and several tech industry leaders called on the U.S. government to support open models and avoid premature restrictions that could lead to innovation outflow.

Businesses can download, use, and modify this new Nemotron model for free, without needing NVIDIA's permission or paying any fees.

For NVIDIA, open-source AI helps drive chip sales: these models still rely on GPUs to operate; at the same time, the lower usage threshold can attract more users, diverting the market for closed-source proprietary models from companies like OpenAI and Anthropic.

Last month, Huang told Axios: "Free and open artificial intelligence is greatly beneficial for hardware and the chip industry as well."

With China's Moonshot AI launching the Kimi K3 model, significantly narrowing the gap with the top U.S. large models, intense discussions erupted in Washington. Huang actively entered this debate. Some politicians are concerned that Kimi K3 may pose national security risks and question whether manufacturers are using knowledge distillation technology to steal intellectual property — this technology trains lightweight models by learning from the outputs of advanced AI services. Policymakers are discussing whether to impose sanctions and various restrictions on Chinese large model companies, similar to controlling chip exports.

This new Nemotron product is available for businesses to download, use, and modify for free, without NVIDIA authorization or payment.

Huang wrote in an open letter: **Open weight models allow businesses to take control of their own development, promote market competition, and reduce usage costs. Open models help enhance security and network protection capabilities, accelerate the diffusion of innovation, and achieve technological autonomy and control **

A few days after the announcement, NVIDIA, along with Microsoft and other companies, established an AI Safety Alliance, focusing on building cybersecurity solutions based on open models and open-source software.

In recent years, retail innovations targeting the pre-teen demographic have been on the rise. Brands like Sephora have witnessed an explosion in demand for children's beauty products; clothing retailers are also launching more exclusive sizes tailored for pre-teens.

At the same time, the Alpha Generation is reshaping the retail landscape. This group of native internet generation consumers is becoming increasingly important, with major brands vying to attract this demographic and cultivate user loyalty. According to the definition by Merriam-Webster, the Alpha Generation roughly refers to those born between 2010 and 2025, although the boundaries of generational classification remain controversial.

Jensen Huang stated that Frida is closely monitoring this trend.

"When parents no longer purchase diapers for infants and toddlers, the subsequent process of selecting related products becomes very fragmented. There is a lack of a one-stop solution that covers all personal care needs for children in this age group, which is precisely where we are focusing our efforts. The same logic applies to our earlier layout in the maternal and infant care sector."

NVIDIA revealed that companies like CrowdStrike, CodeRabbit, and Harvey have completed testing and customization of this latest model. The chip manufacturer stated that Nemotron 3.5 Lightning is designed specifically for AI agents, which are artificial intelligence programs that can operate autonomously in the background.

Nemotron 3.5 Lightning will be available on the HuggingFace platform and NVIDIA's official website. In addition, NVIDIA is simultaneously launching the software tool NeMo Switchyard, which can automatically select the most cost-effective and compatible AI model for any task.

A representative from NVIDIA stated on Monday that the company uses knowledge distillation technology to enable Nemotron 3.5 Lightning to achieve capabilities similar to those of the larger-scale Nemotron series models