![[0c0eb448-5ad6-48fe-9080-3baa40f3d378.mp3]]

DeepMind发布Gemini2.5Flash-Lite：实现实时交互界面生成

简报：

• DeepMind近日推出全新Gemini2.5Flash-Lite模型，该模型可根据用户操作实时解析需求和上下文，动态生成下一屏幕的界面代码和内容，实现“即点即得”的交互体验。
• Gemini2.5Flash-Lite大幅提升了互动自然度和流畅度，用户在应用中无需等待加载或切换，显著增强个性化体验和操作灵活性。
• 相较前代，Gemini2.5Flash-Lite在编码、数学、科学研究与推理等多项任务中表现更优，拓展了其在UI生成、教育、医疗等复杂应用场景的适用性。
• 该技术有望引领人机交互创新，推动更多高度智能化解决方案的发展。

相关链接：

• DeepMind发布Gemini2.5Flash-Lite模型，实现界面实时生成
• DeepMind Gemini2.5Flash-Lite使UI交互更加自然流畅

DeepMind Launches Gemini 2.5 Flash-Lite: Real-time Interactive UI Generation Achieved

Briefing:

• DeepMind recently unveiled its new Gemini 2.5 Flash-Lite model. This model can interpret user needs and context in real-time based on their actions, dynamically generating UI code and content for the next screen, thus enabling a "click-and-get" interactive experience. 🚀
• Gemini 2.5 Flash-Lite significantly enhances interactive naturalness and fluidity. Users no longer need to wait for loading or switching within applications, substantially improving personalized experiences and operational flexibility. ✨
• Compared to its predecessors, Gemini 2.5 Flash-Lite performs better in various tasks including coding, mathematics, scientific research, and reasoning. This expands its applicability in complex scenarios such as UI generation, education, and healthcare. 💡
• This technology is expected to lead innovations in human-computer interaction, driving the development of more highly intelligent solutions.

Related Links:

• DeepMind Unveils Gemini 2.5 Flash-Lite Model for Real-time Interface Generation
• DeepMind Gemini 2.5 Flash-Lite Makes UI Interaction More Natural and Fluid

OpenAI 65亿美元收购Jony Ive公司IO引发与IYO间智能耳塞技术与商标纠纷

简报：

• OpenAI以约65亿美元全资收购苹果前设计总监Jony Ive创建的AI硬件初创公司IO，旨在开发无屏幕AI设备，成为其最大规模的收购；
• 初创公司IYO指控OpenAI与Jony Ive涉嫌抄袭其AI驱动的智能耳塞技术，并指控其新品牌“IO”名称及产品存在知识产权和商标侵权行为；
• IYO已于2025年6月9日正式起诉OpenAI一方，导致OpenAI被法院要求下架与IO相关的全部宣传材料；
• 双方法庭文件披露OpenAI曾对IYO科技及其耳部扫描定制工艺表现出兴趣，但IO高管否认有意推出定制耳塞产品，并称其AI新硬件将不是可穿戴设备；
• 该诉讼正影响着OpenAI在AI硬件领域的声誉，也反映AI设备市场竞争加剧和技术及商标归属争议升温。

相关链接：

• IT之家原文报道
• AIbase详细报道
• OpenAI收购Jony Ive公司相关新闻（MSN）

OpenAI's $6.5 Billion Acquisition of Jony Ive's IO Triggers Smart Earbud Technology and Trademark Dispute with IYO 💰

Briefing:

• OpenAI has fully acquired IO, the AI hardware startup founded by Apple's former design chief Jony Ive, for approximately $6.5 billion. This marks OpenAI's largest acquisition to date and aims to develop screenless AI devices. 🚀
• Startup IYO accuses OpenAI and Jony Ive of allegedly copying its AI-powered smart earbud technology, and claims intellectual property and trademark infringement concerning the new "IO" brand name and its products.
• IYO officially filed a lawsuit against OpenAI on June 9, 2025, leading to a court order for OpenAI to remove all promotional materials related to IO.
• Court documents from both sides reveal that OpenAI had previously shown interest in IYO's technology and its ear-scanning customization process. However, IO executives deny any intention to launch custom earbud products, stating that their new AI hardware will not be a wearable device.
• This lawsuit is impacting OpenAI's reputation in the AI hardware sector, also reflecting the intensifying competition and escalating disputes over technology and trademark ownership in the AI device market. ⚖️

Related Links:

• IT Home Original Report
• AIbase Detailed Report
• OpenAI Acquisition of Jony Ive's Company (MSN)

微软发布3.3亿参数本地AI模型Mu，提升Windows 11自然语言设置体验

简报：

• 微软正式推出本地运行的小型语言模型Mu，拥有3.3亿参数，专为Windows 11设置应用打造AI代理。
• Mu基于编码器-解码器架构，深度适配NPU硬件，在设备端实现低延迟、高隐私的自然语言交互，支持用户通过语音或文本直接控制系统设置。
• 该模型响应速度超每秒200次样本生成，首词生成延迟降低约47%，解码速度提升4.7倍，并通过量化及参数共享等技术优化了内存与算力消耗。
• Mu目前支持搭载高通骁龙X系列处理器的Copilot+ PC，并计划未来扩展至AMD和英特尔平台，推动Windows与AI深度融合。

相关链接：

• 微软发布设备端AI模型 Mu，助力 Win11 系统设置自然语言搜索
• 微软在 Windows 11 推出 3.3 亿参数本地模型Mu
• 微软创新小模型 Mu 赋能 Windows 智能体

Microsoft Unveils Mu: A 330 Million Parameter Local AI Model Enhancing Windows 11 Natural Language Settings Experience ✨

Briefing:

• Microsoft officially launches Mu, a small language model designed to run locally, featuring 330 million parameters, specifically built as an AI agent for the Windows 11 Settings app. 🚀
• Based on an encoder-decoder architecture, Mu is deeply optimized for NPU hardware, enabling low-latency, high-privacy natural language interaction directly on-device. This allows users to control system settings via voice or text.
• The model boasts a response speed exceeding 200 samples per second, reduces first-token generation latency by approximately 47%, and boosts decoding speed by 4.7 times. Memory and computational consumption are further optimized through techniques like quantization and parameter sharing.
• Mu currently supports Copilot+ PCs equipped with Qualcomm Snapdragon X series processors and plans to expand to AMD and Intel platforms in the future, driving deeper integration of Windows with AI. 🤖

Related Links:

• Microsoft Releases On-Device AI Model Mu, Boosting Natural Language Search in Win11 System Settings
• Microsoft Introduces 330 Million Parameter Local Model Mu in Windows 11
• Microsoft's Innovative Small Model Mu Empowers Windows AI Agent

罗博特科子公司ficonTEC签超1亿元光电子封测设备大单

简报：

• 罗博特科公告称，其全资子公司ficonTEC Service GmbH和ficonTEC Automation GmbH与美国某头部公司及其子公司签订了合同，总金额约1710万欧元（折合人民币已超1亿元），占2024年度营业收入超12.82%。
• 本次合同涉及光电子封测设备，顺利履行有助于罗博特科提升相关技术水平及市场竞争力，并预计对本年度及未来业绩产生积极影响。

相关链接：

• 36氪新闻原文

RBT Subsidiary ficonTEC Secures Over 100 Million RMB Photonics Packaging & Testing Equipment Order

Brief:

• RBT announced that its wholly-owned subsidiaries, ficonTEC Service GmbH and ficonTEC Automation GmbH, have signed contracts with a leading U.S. company and its subsidiaries. The total contract value is approximately 17.1 million Euros (exceeding 100 million RMB), representing over 12.82% of the company's 2024 annual operating revenue. 💰
• The contracts involve optoelectronic packaging and testing equipment. Successful fulfillment is expected to help RBT enhance its relevant technological capabilities and market competitiveness, and is projected to have a positive impact on this year's and future performance. ✨🚀

Related Link:

• 36Kr News Original

华为云发布新一代CloudDevice云终端产品，推动智能应用创新

简报：

• 2025年6月21日，华为云在开发者大会（HDC 2025）高峰论坛上正式发布新一代云终端产品CloudDevice，覆盖云手机、云PC、云Pad等形态。
• 华为云联合中国科学院科技战略咨询研究院、中国信通院发布了《云终端产业发展报告》与《基于云网端边芯的云终端技术标准》，CloudDevice通过“云网端边芯”全栈协同，达到4K 90FPS超清画质、端到端60ms超低时延，并获得信通院双项“引领级”认证。
• 华为云表示将持续推动CloudDevice构建开放生态，加速行业智能应用创新，助力智能万物上云。

相关链接：

• 36氪：华为云正式发布新一代CloudDevice
• AiBase: 华为云CloudDevice重磅发布
• AiBase: CloudDevice实现4K超清画质
• AiBase: 云终端产业发展报告发布
• IT之家：华为云发布新一代CloudDevice云终端

Huawei Cloud Launches Next-Gen CloudDevice Cloud Terminal Products, Driving Smart Application Innovation

Brief:

• On June 21, 2025, Huawei Cloud officially launched its next-generation CloudDevice cloud terminal products, including cloud phones, cloud PCs, and cloud Pads, at the Huawei Developer Conference (HDC 2025) Summit Forum.
• Huawei Cloud, in collaboration with the Institutes of Science and Development (ISD) of the Chinese Academy of Sciences and the China Academy of Information and Communications Technology (CAICT), released the 'Cloud Terminal Industry Development Report' and the 'Cloud Terminal Technical Standard Based on Cloud-Network-Device-Edge-Chip Synergy'. CloudDevice achieves 4K 90FPS ultra-clear picture quality and end-to-end ultra-low latency of 60ms through full-stack synergy across 'Cloud-Network-Device-Edge-Chip', and has received dual 'Leading Level' certifications from CAICT. 🚀
• Huawei Cloud stated it will continue to promote CloudDevice to build an open ecosystem, accelerate innovation in smart industry applications, and help connect everything intelligent to the cloud. ☁️

Related Links:

• 36Kr: Huawei Cloud Officially Launches Next-Gen CloudDevice
• AiBase: Huawei Cloud CloudDevice Grand Launch
• AiBase: CloudDevice Achieves 4K Ultra-Clear Picture Quality
• AiBase: Cloud Terminal Industry Development Report Released
• IT Home: Huawei Cloud Releases Next-Gen CloudDevice Cloud Terminal 💡