TLDR
- Nvidia launched third-generation Nemotron open-source AI models, with Nemotron 3 Nano available now and two larger versions coming in first half of 2026
- The new models are faster, cheaper to run, and better at handling long, multi-step tasks compared to previous versions
- Release comes as Chinese AI labs like DeepSeek, Moonshot AI, and Alibaba gain traction with their open-source offerings
- Meta Platforms is reportedly considering a shift toward closed-source models, making Nvidia a leading U.S. open-source provider
- Nvidia is openly releasing training data and tools to allow government and business users to test security and customize models
Nvidia rolled out its third-generation Nemotron large-language models on Monday. The release puts the chip maker deeper into the open-source AI software game.
NVIDIA Corporation, NVDA
The smallest model, Nemotron 3 Nano, became available immediately. Two larger versions will launch during the first half of 2026. The models handle writing, coding, and other tasks.
Nvidia claims the new Nemotron 3 Nano runs more efficiently than older versions. This means lower operating costs for users. The model also performs better on lengthy tasks requiring multiple steps.
The company built its reputation selling chips to AI firms like OpenAI. But Nvidia also develops its own open-source models. These cover everything from physics simulations to self-driving vehicles. Companies like Palantir Technologies incorporate these models into their products.
The release comes at an interesting time. Chinese tech companies are flooding the market with open-source AI offerings. DeepSeek, Moonshot AI, and Alibaba Group Holdings have all gained ground in the tech industry.
Airbnb recently disclosed it uses Alibaba’s Qwen open-source model. This shows how Chinese models are spreading through U.S. companies.
Meta’s Potential Shift Changes the Landscape
Reports from CNBC and Bloomberg suggest Meta Platforms might move toward closed-source models. If true, this would leave Nvidia as one of the few major U.S. providers of open-source AI.
Many U.S. states and government agencies have banned Chinese models due to security concerns. This creates an opportunity for Nvidia to fill the gap.
Kari Briski, Nvidia’s vice president of generative AI software for enterprise, emphasized reliability. The company wants to provide “a model that people can depend on.”
Training Data and Security Tools
Nvidia is releasing its training data openly. The company also provides tools for security testing and customization. Government and business users can examine the models thoroughly.
“This is why we’re treating it like a library,” Briski said in an interview with Reuters. “This is why we’re committed to it from a software engineering perspective.”
The move addresses security concerns that have plagued Chinese models. By opening up its development process, Nvidia aims to build trust with enterprise and government customers.
The company has become the world’s most valuable listed company. Its chips power most AI training operations globally.
Nvidia’s open-source strategy differs from competitors who charge for access to their models. The company makes money from chip sales rather than software licensing.
The Nemotron 3 Nano release marks the beginning of Nvidia’s 2026 model rollout. The two larger versions scheduled for next year will likely target more demanding enterprise applications.
The post Nvidia (NVDA) Stock Climbs Higher as Chip Giant Launches New Open-Source AI Models appeared first on Blockonomi.