Jensen Huang just used GTC to reposition Nvidia from a chip company to an open-source AI platform. NemoClaw, open frontier models, Nemotron —he's doing more for open-source AI than any CEO in America. It sounds generous, but it's strategic. Nvidia gives away the software and monetizes what it runs on. Zuckerberg spent years and billions trying to escape his dependence on Apple and Google's platforms. Huang is making sure Nvidia never ends up in that position. Hugging Face CEO Clement Delangue joins live. Hu ...

RT vitalik.eth (@VitalikButerin)Also on this topic, it's worth highlighting how the recent Anthropic news (reminder: in the same week, they (i) refused to allow DoW to use their AI for mass surveillance of Americans and fully autonomous weapons, but also (ii) cancelled their safety pledge, and (iii) argued that China bad for distilling their models and making open-weights models) can be perceived by people who do not have "Anthropic bags" (or "America bags").Here's one fascinating article from China:https:/ ...

Also on this topic, it's worth highlighting how the recent Anthropic news (reminder: in the same week, they (i) refused to allow DoW to use their AI for mass surveillance of Americans and fully autonomous weapons, but also (ii) cancelled their safety pledge, and (iii) argued that China bad for distilling their models and making open-weights models) can be perceived by people who do not have "Anthropic bags" (or "America bags").Here's one fascinating article from China:https://t.co/aXiTq6QDooExcerpt translat ...

Core Insights - The open-source AI assistant has gained significant popularity within China's tech community due to its ability to make and execute decisions on behalf of users [1] Group 1 - The AI assistant operates on an open-source model, allowing for broader accessibility and customization [1] - Its decision-making capabilities have attracted attention, indicating a shift towards more autonomous AI applications in technology [1] - The success of this AI assistant reflects the growing interest and investment in AI technologies within the Chinese market [1]

China’s enthusiasm for OpenClaw is igniting a stock rally in local technology firms moving swiftly to embrace the open-source AI program https://t.co/UvBpj13O8S ...

Indian AI lab Sarvam’s new models are a major bet on the viability of open-source AI https://t.co/kYOqFS6Z41 ...

Group 1 - Alibaba Group Holding Limited (NYSE:BABA) is recognized as one of the best stocks to buy and hold for 2026, focusing on advancements in open-source AI models [1] - The company launched its open-source CosyVoice 3 model on December 17, which is a multilingual speech synthesis model trained on over 1 million hours of audio and supports nine languages [1] - Alibaba also announced a significant upgrade to its AgentScope, enabling the building and deployment of enterprise-grade AI agents with enhanced applications and infrastructure [2] Group 2 - A recent report indicates that Alibaba and other Chinese companies are seeking H200 chips from Nvidia, following approval from the Trump administration for chip exports to China [3] - The Chinese government is currently assessing the demand for H200 chips from these companies, which could expedite the training, launching, and deployment of Alibaba's AI models if approved [3] - Alibaba operates as a multinational tech giant, focusing on e-commerce, retail, cloud computing, and digital services, aiming to facilitate global business connections through platforms like Alibaba.com [4]

Core Insights - Nvidia has announced the Nemotron 3 family of openly released AI models, training datasets, and engineering libraries, indicating a significant move towards open-source AI development [1] Group 1 - The introduction of the Nemotron 3 family represents an aggressive push by Nvidia into the open-source AI sector [1]

Model Performance & Benchmarks - Kimmy K2 Thinking outperforms GPT-5 on the "Humanity's Last Exam" benchmark with a score of 44.9% compared to GPT-5's 41.7% [1] - In agentic search for Browse Comp, Kimmy K2 Thinking scores 60.2% versus 54.9% for GPT-5 [1][2] - Kimmy K2 Thinking achieves 83.1% on Live Codebench v6, a competitive programming benchmark [1] - The model can execute 200 to 300 sequential tool calls without human interference [1][2] - Kimmy K2 Thinking significantly outperforms the human baseline of 29.2% on browse comp with a score of 60.2% [2] Model Architecture & Training - The base Kimmy K2 model used 2.8 million H800 hours with 14.8 trillion tokens, costing approximately $5.6 to $6 million [3] - Kimmy K2 Thinking has a trillion parameters with 384 experts, while 32 billion parameters are active during inference [5][6] - Kimmy K2 Thinking has a vocabulary size of 160,000 [5] Market & Industry Impact - China is emerging as a key player in open-source, open-weights frontier AI models [9][10] - The cost of training frontier models is decreasing rapidly [3][4] Use Cases & Capabilities - Kimmy K2 Thinking can solve PhD-level mathematics problems using 23 tool calls in its chain of thought [1] - The model can create component-heavy websites and math explainer visualizations from single prompts [1] - Kimmy K2 Thinking can analyze the relationship between population density and healthcare facility accessibility, generating interactive maps and charts [11][12][13][14][15]

Group 1 - ASML and Mistral AI have formed a partnership to explore AI models across ASML's lithography systems, with ASML becoming an 11% shareholder in Mistral AI [1][2] - The partnership symbolizes the importance of sovereign AI and trade barriers, bringing together Mistral AI and ASML, a key player in global technology [2] - Mistral AI has received significant US funding and was previously considered for acquisition by major companies like Microsoft, but ASML's investment helps maintain its independence [2] Group 2 - Mistral AI differentiates itself by leveraging open-source models and targeting enterprise customers, launching its first reasoning model, Magistral, in June [3] - The company employs innovative mixture of experts (MoE) architectures, achieving high performance per compute cost, making its models cost-effective [4] - In February 2024, Mistral AI launched Le Chat, a multilingual conversational assistant, enhancing its enterprise platform with free advanced features and extensive third-party integrations [5]