Nvidia launches Nemotron 3 open source AI models as Meta steps back from transparency
16 Sources
[1]
Nvidia Becomes a Major Model Maker With Nemotron 3
Nvidia has made a fortune supplying chips to companies working on artificial intelligence, but today the chipmaker took a step toward becoming a more serious model maker itself by releasing a series of cutting-edge open models, along with data and tools to help engineers use them. The move, which
[2]
As Meta fades in open-source AI, Nvidia senses its chance to lead
Nvidia argues it's more open than Meta with data transparency. Seizing upon a shift in the field of open-source artificial intelligence, chip giant Nvidia, whose processors dominate AI, has unveiled the third generation of its Nemotron family of open-source large language models. The new Nemotron
[3]
Not enough good American open models? Nvidia wants to help
Nemotron 3 is a grab bag of 2025's top machine learning advancements For many, enterprise AI adoption depends on the availability of high-quality open-weights models. Exposing sensitive customer data or hard-fought intellectual property to APIs so you can use closed models like ChatGPT is a
[4]
Nvidia bets on open infrastructure for the agentic AI era with Nemotron 3
The company is positioning its new offerings as a business-ready way for enterprises to build domain-specific agents without first needing to create foundation models. AI agents must be able to cooperate, coordinate, and execute across large contexts and long time periods, and this, says Nvidia,
[5]
Nvidia Wants To Be Your Open Source Model Provider
It seems that chips dominance isn't enough for Nvidia. The tech giant wants its products to be present at every level of the AI development ecosystem, and its got its eyes on open-source AI models. Nvidia released its latest Nemotron 3 family of open models on Monday, only months after announcing
[6]
From Slurm to Nemotron 3, Nvidia unveils big open-source push
Nvidia launched Nemotron 3 models, including Nano, Super, and Ultra sizes for AI tasks Nvidia has announced a major expansion of its open source efforts, combining a software acquisition with new open AI models. The company has announced it has acquired SchedMD, the developer of Slurm, an open
[7]
Nvidia debuts Nemotron 3 with hybrid MoE and Mamba-Transformer to drive efficient agentic AI
Nvidia launched the new version of its frontier models, Nemotron 3, by leaning in on a model architecture that the world's most valuable company said offers more accuracy and reliability for agents. Nemotron 3 will be available in three sizes: Nemotron 3 Nano with 30B parameters, mainly for
[8]
Analysis: Nvidia Nemotron-3 open models lead to more efficient agentic AI - SiliconANGLE
Analysis: Nvidia Nemotron-3 open models lead to more efficient agentic AI Artificial intelligence leader Nvidia Corp. Monday announced the Nemotron-3 family of models, data and tools, and the release is further evidence of the company's commitment to the open ecosystem, focusing on delivering
[9]
NVIDIA Unveils Nemotron 3 Open Models to Power Multi-Agent AI Systems | AIM
The Nemotron 3 lineup includes Nano, Super and Ultra models built on a hybrid latent mixture-of-experts (MoE) architecture. NVIDIA on Monday announced the NVIDIA Nemotron 3 family of open models, datasets and libraries aimed at supporting the development of transparent and efficient multi-agent AI
[10]
Nvidia launches Nemotron 3 model family as open foundation for agentic AI systems - SiliconANGLE
Nvidia launches Nemotron 3 model family as open foundation for agentic AI systems Nvidia Corp. today announced the launch of Nemotron 3, a family of open models and data libraries aimed at powering the next generation of agentic artificial intelligence operations across industries. The new family
[11]
NVIDIA debuts Nemotron 3 open models to accelerate agentic AI development
The Nemotron 3 line of open models has been released by NVIDIA, focusing on providing transparent, efficient, and highly accurate AI functionality for building collaborative agentic systems. The Nemotron 3 line comes in different sizes, including Nano, Super, and Ultra, and boasts architectural
[12]
AI Chipmaker Nvidia Just Released Open Source Models For AI agents - NVIDIA (NASDAQ:NVDA)
Nvidia on Monday introduced the Nemotron 3 family of open models, data and libraries designed to power and support open, efficient, specialized AI agents. The company said these models help teams build reliable AI systems that work together. At the same time, Nvidia is also expanding its open
[13]
Nvidia Bets on Open Models to Power AI Agents | PYMNTS.com
By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. The lineup includes Nano, Super and Ultra models that leverage a hybrid latent
[14]
Nvidia unveils new open-source AI models amid boom in Chinese offerings
SAN FRANCISCO -- Nvidia on Monday unveiled a new family of open-source artificial intelligence 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 primarily known for providing chips that firms
[15]
Nvidia launches new open-source AI models amid rise of Chinese alternatives
On Monday Nvidia unveiled a new generation of open-source artificial intelligence models, marking a strategic offensive in response to the rise of models developed by Chinese labs such as DeepSeek, Alibaba and Moonshot AI. The American group, known for its chips essential to training proprietary
[16]
NVIDIA powers the agentic era: The Nemotron 3 debut explained
Open weight NVIDIA Nemotron models drive sovereign AI agentic systems The era of the solitary chatbot is fading. As 2025 draws to a close, the artificial intelligence industry is pivoting aggressively toward "agentic AI" - systems where multiple specialized AI agents collaborate to solve complex
Share
Copy Link
Nvidia released its Nemotron 3 family of models in three sizes—Nano (30B), Super (100B), and Ultra (500B parameters)—with full training data transparency. The move positions the chip giant as a major force in open source AI just as Meta signals a retreat from openness, leaving Chinese models to dominate the space.
Nvidia Takes Leadership Position in Open Source AI
Nvidia has stepped decisively into the open source AI arena with the release of its Nemotron 3 family of models, positioning itself as a transparency leader at a moment when American tech giants are retreating from openness
1
. The chip giant, which has built its fortune supplying GPUs to AI companies, is now hedging against the possibility that firms like OpenAI, Google, and Anthropic might develop their own chips and move away from Nvidia's technology over time1
.The Nemotron 3 family of models arrives in three sizes: Nano with 30 billion parameters, Super with 100 billion, and Ultra with 500 billion
1
. The Nano model, available now on Hugging Face, increases throughput in tokens per second by four times and extends the context window to one million tokens—seven times larger than its predecessor2
. Super is expected in January, while Ultra will arrive in March or April2
.
Source: Digit
Enterprise AI Adoption Drives Model Customization Focus
Nvidia is taking a more transparent approach than many US rivals by releasing the training data used to train Nemotron, a fact that should help engineers with model customization more easily
1
. The company is also releasing tools including a new hybrid latent mixture-of-experts model architecture, which Nvidia says excels at building AI agents that can take actions on computers or the web1
.Kari Briski, vice president of generative AI software for enterprise at Nvidia, explained that open AI models matter to builders for three key reasons: the growing need to customize models for particular tasks, the benefit of routing queries to different models, and the ability to extract more intelligent responses through simulated reasoning
1
. For enterprise AI adoption, exposing sensitive customer data or intellectual property to APIs for closed models like ChatGPT remains a non-starter3
.The company is releasing reinforcement learning libraries and training environments called NeMo Gym, enabling users to train AI agents through trial and error with rewards and punishments
3
. Jensen Huang stated that "open innovation is the foundation of AI progress," emphasizing how Nemotron transforms advanced AI into an open platform that gives developers the transparency and efficiency needed to build agentic systems at scale1
.
Source: SiliconANGLE
Meta's Retreat Creates Opening in AI Development Ecosystem
The timing of Nvidia's push is strategic. Meta, which released the first advanced open models under the name Llama in February 2023, has signaled that future releases might not be open source
1
. Meta's rollout of Llama 4 in April 2025 received mediocre reviews, and Llama models no longer appear in the top 100 on LMSYS's LMArena Leaderboard, which is dominated by proprietary models from Google, xAI, Anthropic, and OpenAI, along with open source AI models from DeepSeek AI, Alibaba's Qwen, and Singapore-based Moonshot AI2
.A forthcoming Meta project code-named Avocado may launch as a closed model that Meta can sell access to, marking the biggest departure from the open source strategy Meta has promoted for years
2
. Meta's Chief AI Officer Alexandr Wang, installed this year, is an advocate of closed models2
. When asked about Menlo Ventures' claim that open source is struggling, Briski replied: "I agree about the decline of Llama, but I don't agree with the decline of open source"2
.Chinese AI Models Dominate as American Firms Turn Proprietary
With American tech giants drawing back from open source, Chinese AI models have dominated the space with high-profile releases from DeepSeek and Alibaba
5
. Open models from Chinese companies are currently much more popular than those from US firms, according to data from Hugging Face1
. Even OpenAI, which purports to be an "open" organization, unveiled only its first open source model in five years this past August with gpt-oss-120b and gpt-oss-20b5
.At Nvidia's GTC conference in Washington DC last October, Jensen Huang warned attendees of China's open source dominance, arguing that because China creates substantial open source software, Americans risk being "ill-prepared" if Chinese software "permeates the world"
5
. Briski emphasized that "this year alone, we had the most contributions and repositories on Hugging Face," signaling Nvidia's growing prominence2
.Related Stories
Technical Architecture Targets AI Infrastructure Needs
The Nemotron 3 models employ a novel hybrid latent mixture-of-experts architecture designed to minimize performance losses when processing long input sequences
3
. This combines Mamba-2 and Transformer architectures throughout the model's layers, with Mamba-2 providing efficiency for long sequences while transformers maintain precise reasoning and prevent context loss3
. All models natively support a million token context window, equivalent to roughly 3,000 double-spaced pages3
.The mixture-of-experts architecture means only a fraction of total parameters activates for each token processed. Nemotron 3 Nano has 30 billion parameters but only 3 billion activate per token, putting less pressure on memory and delivering faster throughput than equivalent dense models
3
. The larger Super and Ultra models use Nvidia's NVFP4 data type and a latent MoE architecture where experts operate on shared latent representation, allowing models to call on 4x more experts at the same inference cost3
.Strategic Positioning for Hardware Dominance
Nvidia sees Nemotron as a tool to drive further reliance on its hardware empire. By distributing open models to developers worldwide, the company aims to ensure that new foundation models built will align best with its silicon rather than rapidly emerging Chinese alternatives
5
. "When we're the best development platform, people are going to choose us, choose our platform, choose our GPU, in order to be able to not just build today, but also build for tomorrow," Briski said5
.The 30-billion-parameter Nemotron 3 Nano runs efficiently on enterprise hardware like Nvidia's L40S or RTX Pro 6000 Server Edition, though 4-bit quantized versions should work on GPUs with as little as 24GB of video memory
3
. According to Artificial Analysis, the model delivers performance on par with gpt-oss-20B or Qwen3 models while offering greater flexibility for customization3
. Nvidia is positioning these offerings as business-ready infrastructure for enterprises to build domain-specific AI agents without needing to create foundation models from scratch4
.
Source: The Register
References
Summarized by
Navi
[3]
Related Stories
Nvidia launches Nemotron 3 Super with 5x throughput boost to power enterprise AI agents
11 Mar 2026•Technology
Nvidia releases Nemotron 3.5 Lightning and NeMo Switchyard to slash agentic AI costs by 66%
11 Aug 2026•Technology
NVIDIA Unveils Nemotron Model Families to Advance AI Agents and Enterprise Productivity
07 Jan 2025•Technology
Recent Highlights
1
AI Hallucination Nearly Sparked US Military Confrontation With Chinese Ship in Middle East
Technology
2
OpenAI AI Agent Hacks Australian Government Healthcare Website in First-of-Its-Kind Breach
Technology
3
AI leaders warn UN Security Council of existential risks to humanity, call for global safeguards
Policy and Regulation
Recent Highlights
Today's Top Stories
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
