TORONTO, ON / ACCESS Newswire / January 8, 2026 / Datametrex AI Limited (the "Company" or "Datametrex") (TSXV:DM)(FSE:D4G)(OTC:DTMXF) is pleased to announce that it has received new Purchase Orders ("P.O.s") with a combined value of approximately CAD $400,000. These P.O.s were secured through the Company's wholly owned subsidiary, Datametrex Korea, and represent continued progress in the execution of Datametrex's global growth strategy. ...

Group 1 - Moole Technology released version 1.1 of the open-source large model distributed training simulation tool SimuMax [1] - Alibaba Cloud launched a multimodal interaction development kit aimed at AI hardware [1] - Xiaopeng Motors Chairman He Xiaopeng announced plans for mass production of humanoid robots and flying cars this year [1] Group 2 - JD.com established a "Chameleon Business Unit" to undertake AI initiatives [1] - Tesla CEO Elon Musk stated that within three years, robot surgeons will surpass top human doctors, and China's AI computing power will lead the world [1] - Caterpillar partnered with NVIDIA to transform heavy industry through physical AI and robotics technology [1] Group 3 - Anthropic, a rising star in artificial intelligence, plans to raise $10 billion, with its valuation nearing that of OpenAI [1] - Arm restructured the company and established a physical AI department [1] - Datavault AI expanded its collaboration with IBM [1] - Tensu Zhixin will unveil the roadmap for its next three generations of GPGPU on January 26 [1] - Leju Robotics formed a full-stack AI partnership with Alibaba Cloud [1]

#ChatGPT发布美国版蚂蚁阿福# 刚刚，ChatGPT 发布 AI 医疗功能！能读病历做诊断，支持苹果健康 2026 年，OpenAI 决定认真入局医疗健康领域了。 就在刚刚，OpenAI 重磅推出了 ChatGPT 健康（ChatGPT Health），能够把你的医疗记录、健康 App、 甚至是 Apple 健康数据全给串起来，然后用 AI 帮你看懂那些复杂到头大的体检报告、准备就医问题清 单、甚至规划饮食运动。 根据 OpenAI 自己透露的数据，健康咨询已经成了 ChatGPT 最高频的使用场景之一，全球每周有超过 2.3 亿人在上面问健康问题。于是，OpenAI 索性直接做个专门的健康版本。 当然，OpenAI 也强调，ChatGPT 健康只是辅助工具，不能替代医生看病，这话既是「免责金牌」，也 是 AI 医疗绕不开的「紧箍咒」，更专业的诊断，还得靠临床医生拍板。 ...

Core Insights - Tuya Smart hosted "Tuya Developer Day" at CES 2026, focusing on the theme "Accelerating the Physical AI Era Together," gathering leaders from top tech companies to discuss AI advancements [1] Leadership Dialogue - Alex Yang, COO and CFO of Tuya, discussed the transition of AI from concepts to large-scale applications with representatives from Google Cloud and AWS, emphasizing insights on AI-native devices [2] - Praveen Rao from Google Cloud highlighted the evolution of multimodal capabilities in AI systems, indicating a shift from virtual applications to real-world integration, aiming for proactive AI services [3] - Danny Smith from AWS noted the development of enterprise-scale agentic AI with Tuya, which can process multidimensional data and make optimal decisions rapidly, enhancing business outcomes [3] Panel Discussion - Chuck Sabin from Bluetooth SIG emphasized the importance of connectivity as a driver of innovation, collaborating with Tuya to enhance interoperability across devices [4] - Jeff Platon from the Wi-Fi Alliance discussed how AI is transforming connectivity experiences, ensuring Wi-Fi remains a secure foundation for intelligent applications [4] - Jon Harros from the Connectivity Standards Alliance stressed the need for open standards and collaboration to foster secure, interoperable AI solutions, with Tuya playing a key role [4] - Colin Cureton from Silicon Labs pointed out the challenges of fragmentation in deploying AI-native systems, advocating for partnerships to create robust connectivity platforms [4] Keynote Presentations - Alex Yang presented on how AI is redefining life, focusing on creating intuitive experiences that integrate smart devices and services to enhance daily living [5] - Joe Sun, CEO of Robopoet, discussed emotional companion robots for Gen-Z, highlighting their ability to form deeper emotional connections through Tuya's technology [5] - Olivier Garreau from Wisdom Oasis AI emphasized AI's role in enhancing parent-infant interactions, showcasing Tuya's technology in understanding subtle needs [5] - The event showcased breakthroughs in AI technology, including multimodal perception and emotional companionship, with a commitment to advancing Physical AI through collaboration [5]

技术路线以MoE架构为核心，早于多数同行布局多模态并行研发 公司全栈自研，研发效率高，截至2025年9月累计投入研发约5亿美元。2025年迭代多个模型：Speech系 列在语音生成领域领先，Hailuo系列视频模型在全球基准测试中位居前列，开源模型MiniMax-M1和M2 在Artificial Analysis榜单表现突出。竞争优势主要体现在三个方面。第一，全球化执行力强。Talkie在海 外AI伴侣赛道领先，海螺AI视频生成用量大，海外市场付费意愿高，避免价格战。第二，双轮模式均 衡。消费者业务提供规模和数据，企业业务保障高毛利和稳定性，抗风险能力较好。第三，研发效率领 先。团队年轻化、扁平化管理，结合MoE和自研基础设施，单位算力产出高。 东吴证券发布研报称，MINIMAX-WP(00100)成立于2021年12月，专注通用人工智能基础技术研发，目 标是构建国际竞争力的多模态大模型，并面向全球市场提供产品和服务。从成立之初，公司就选择全球 化定位，同时布局消费者产品和海外渠道。2025/12/21通过港交所聆讯，12月31日启动港股IPO，股票 代码0100，发行估值461-504亿港元，计划募资超过 ...

Core Insights - Datavault AI's stock rose by 4.76% in pre-market trading [1] - The company is expanding its collaboration with IBM [1] - Datavault AI will utilize IBM's Available Infrastructure and SanQtum AI platform to deploy AI solutions at the enterprise edge [1]

Jay 发自 凹非寺 量子位 | 公众号 QbitAI 盼星星盼月亮，千呼万唤的DeepSeek-R2没盼到， R1又抱着琵琶出来溜了一圈 。 还记得去年登上《Nature》封面的那篇关于R1的论文吗？ DeepSeek又悄悄给它塞了 64页 的技术细节。 是的，你没看错，直接从22页干到86页，简直可以当教科书看了…… 谁能想到，论文发布都快一年过去了，DeepSeek居然还能更这么多东西。 DeepSeek怒加64页猛料 把两份论文对着一看，发现这件事不简单。 新版本论文的信息量很大，不止是补了几块附录，正文也被大幅度翻修，几乎像重写了一篇。 在看新论文前，先简单回溯下去年一月份发的v1版。 这个版本围着DeepSeek-R1-Zero展开，重点是释放信号：纯强化学习这条路，是能走通的。 相比起来，v2明显在具体的实现细节上下了更多笔墨。 就比如R1这部分，DeepSeek这次系统性把R1的完整训练路径展开了。 整个过程分成四步： 第一步，冷启动 。用数千条能体现思考过程的CoT数据，对模型做SFT。 第二步，推理导向RL 。在不破坏对话思考风格的前提下，继续提升模型能力，同时引入语言一致性奖励，解决语种 ...

Core Viewpoint - The acquisition of startup Manus for $2.5 billion raises concerns in Beijing about encouraging more entrepreneurs to pursue similar paths [1] Group 1 - The $2.5 billion acquisition is seen as a significant move that may set a precedent for other startups [1] - Beijing's worries stem from the potential increase in entrepreneurial activities that could follow this acquisition [1]

（来源：CSHIA） 2026年1月，聆思正式上线 LNN（Listenai Neural Network）算法训练推理平台 3.0。 作为全自主研发的轻量级神经网络推理框架，LNN自2023年10月在GitHub开源以来，凭借其 "低代码甚至零代码" 的AI部署理念，已广泛应用于端侧AI场 景。 全面拓展硬件支持 覆盖全系芯片平台 LNN 3.0显著增强了对聆思自研芯片生态的支持。在原有支持Venus系列的基础上，新增对Arcs、VenusA及Mars全系列芯片平台的完整兼容： 本次3.0版本的全面升级，聚焦硬件兼容性扩展、量化训练性能跃升、推理工具链智能化三大方向，标志着LNN从"可用"迈向"好用、高效、专业"的新阶 段。 聆思始终致力于打通"芯片—算法—大模型—平台工具"的完整技术闭环，为行业提供友好、高效、一站式端侧AI开发体验。 此次LNN3.0的发布，不仅强化平台自身的易用性与性能，更将为聆思全栈生态体系注入更强的开发动能，进一步推动行业加速实现端侧 AI 的规模化落地 与创新应用。 - 聆思AI生态框架图 - 1 新增 Arcs、VenusA 、Mars全系列支持； VenusA 是25年新发布 ...

据悉，其前瞻性的MaaS平台，已吸引超过270万企业及开发者接入，覆盖中国前十大互联网公司中的九家。其代码生成产品Coding在短时间内实现年度经 常性收入（ARR）破亿，成为国内开发者工具的标杆。 作者 | 峰会组委会 摘要：智谱登陆港交所，标志着以大模型为核心的"十五五"新产业爆发前夜已经到来。值此历史性时刻，于2026年1月21日至22日举办的第十届（2026） 中国不动产资产管理峰会暨第四届智慧低碳产融峰会，旨在系统挖掘并把握新产业全生命周期的价值机遇。 2026年伊始，人工智能产业迎来标志性事件——被誉为"中国版OpenAI"的智谱公司正式登陆港交所，股票代码"02513.HK"，一举成为全球首家AGI基座模 型上市公司。上市首日，智谱开盘即涨超3%，报120港元/股，市值突破528亿港元，为AI资本市场拉开序幕。 智谱登陆港交所，昭示着大模型驱动的"十五五"新产业爆发前夜。值此历史性时刻，于2026年1月21日至22日举办的第十届（2026）中国不动产资产管理 峰会暨第四届智慧低碳产融峰会，旨在系统挖掘并把握新产业全生命周期的价值机遇。 技术领跑全球，商业化路径清晰 智谱此次IPO获得超额认购11 ...