单词学习：![[5ba98aad-fd5e-475d-9015-3c1aff2135fd.mp3]]

新闻播报：![[ab70ec01-b330-4ad1-8ae6-b0d512897354.mp3]]

钉钉AI升级：聊天可智能回复，会议录音能问答

简报：

• 钉钉发布8.1.10版本，将AI功能深度集成到日常沟通中，覆盖消息、会议和硬件设备。
• 聊天场景新增“AI灵动回复”，可根据上下文智能生成回复建议；同时提供AI表情表态和转发消息自动总结功能。
• 会议与录音场景中，AI听记功能升级，支持基于录音内容进行问答、提取待办事项和生成总结。
• 硬件方面，DingTalk A1录音设备新增免费的多语种实时翻译功能，可作为同声传译机使用，并优化了文件传输与管理体验。

相关链接：

• 钉钉8.1. 10 重磅升级：AI搬进每条消息、每场会议、每台录音设备

DingTalk AI Upgrade: Chat with Smart Replies, Q&A on Meeting Recordings 🤖

Brief:

• DingTalk has released version 8.1.10, deeply integrating AI functionalities into daily communication, covering messages, meetings, and hardware devices.
• In chat scenarios, a new "AI Dynamic Reply" feature has been added, which can intelligently generate reply suggestions based on context. It also offers AI emoji reactions and automatic summarization for forwarded messages.
• For meetings and recordings, the AI transcription feature has been upgraded, now supporting Q&A based on recorded content, extraction of to-do items, and summary generation. 📝
• In terms of hardware, the DingTalk A1 recording device now includes a free multi-language real-time translation function, allowing it to be used as a simultaneous interpretation machine, and has optimized file transfer and management experiences. 🎤

Related Link:

• DingTalk 8.1.10 Major Upgrade: AI Moves into Every Message, Every Meeting, Every Recording Device

Mistral推出新一代开源编码模型Devstral 2

简报：

• Mistral AI发布专为编码设计的新一代开源模型家族Devstral2，包含123B参数的旗舰版和24B的轻量版。
• 旗舰模型Devstral2在权威软件工程基准测试SWE-bench Verified上获得72.2%的得分，支持256K上下文窗口。
• 同步推出开源命令行工具Mistral Vibe CLI，支持开发者通过自然语言直接操作代码库，实现端到端自动化编程。
• Devstral2模型已通过官方API免费开放使用，并采用修改版MIT及Apache 2.0许可开源。

相关链接：

• Mistral重磅开源Devstral 2：123B编码神器+SWE-bench 72.2分，免费API+本地CLI炸裂来袭！
• 蚂蚁开源 LLaDA2.0，业内首个100B 参数规模的扩散语言模型
• 智谱多模态开源周圆满落幕：四项视频生成核心技术全面开放

Mistral Launches Next-Generation Open-Source Coding Model Devstral 2

Brief:

• Mistral AI has released Devstral 2, a new family of open-source models specifically designed for coding, featuring a flagship version with 123B parameters and a lightweight 24B version. 💻
• The flagship Devstral 2 model achieved a score of 72.2% on the authoritative software engineering benchmark SWE-bench Verified and supports a 256K context window.
• Simultaneously, Mistral Vibe CLI, an open-source command-line tool, has been launched, enabling developers to directly manipulate code repositories via natural language for end-to-end automated programming. 🚀
• The Devstral 2 models are now freely available via official API and are open-sourced under modified MIT and Apache 2.0 licenses. ✨

Related Links:

• Mistral Open-Sources Devstral 2: 123B Coding Powerhouse + SWE-bench 72.2 Score, Free API + Local CLI Launching with a Bang!
• Ant Group Open-Sources LLaDA2.0, the Industry's First Diffusion Language Model with 100B Parameters
• Zhipu AI Multimodal Open Source Week Concludes Successfully: Four Core Video Generation Technologies Fully Open

AI大模型上市潮：三家头部公司竞逐港股首发

简报：据多家媒体援引市场消息，为争夺“中国大模型第一股”的称号，国产AI公司MiniMax、智谱AI及月之暗面正密集筹备赴港上市，其中MiniMax可能最早于2026年1月启动IPO。

相关链接：

• 国产大模型“抢滩”港股：MiniMax、智谱、月之暗面竞逐“AI第一股”

The Great AI Large Model IPO Race: Three Leading Firms Eye Hong Kong Listing 🚀

Brief: According to market sources cited by multiple media outlets, Chinese AI companies MiniMax, Zhipu AI, and Moonshot AI are intensively preparing for their Hong Kong listings in a bid to claim the title of "China's First Large Model Stock," with MiniMax potentially initiating its IPO as early as January 2026. 🗓️

Related Links:

• Chinese Large Models 'Storm' Hong Kong: MiniMax, Zhipu, Moonshot Race for 'First AI Stock' 💰

英伟达召集初创企业，密会商讨AI“电荒”难题

简报：

• 据The Information援引消息人士报道，英伟达将于下周在其加州圣克拉拉总部，针对初创企业举办一场闭门峰会。
• 会议将重点讨论数据中心的供电瓶颈问题，该问题被认为是AI基础设施面临的压力，并可能拖慢AI发展进度。
• 与会者预计为来自电力与电气工程相关初创企业的高管，其中也包括英伟达已投资的公司。

相关链接：

• 应对 AI 时代“缺电”之困，消息称英伟达下周举办闭门峰会

Nvidia Convenes Startups for Secret Summit on AI's "Power Crunch" 🤝

Brief:

• According to sources cited by The Information, Nvidia will host a closed-door summit for startups next week at its Santa Clara, California headquarters.
• The meeting will focus on the power supply bottlenecks facing data centers, a critical issue perceived as putting pressure on AI infrastructure and potentially slowing down AI development. ⚡️
• Attendees are expected to be executives from power and electrical engineering-related startups, including companies that Nvidia has already invested in. 💡

Related link:

• Nvidia to Hold Closed-Door Summit Next Week to Address "Power Shortage" in AI Era

谷歌抢发AI研究代理，与OpenAI展开正面竞争

简报：

• 在OpenAI发布GPT-5.2前夕，谷歌抢先推出基于Gemini 3 Pro模型的AI研究代理——Gemini Deep Research。
• 该工具专为处理海量信息和复杂研究任务设计，旨在提升事实准确性，并将通过新的Interactions API向开发者开放。
• 谷歌计划将Gemini Deep Research集成到Google Search、Gemini App及NotebookLM等产品中，加速布局“代理式AI”时代。
• 谷歌发布后，OpenAI紧随其后正式推出GPT-5.2并宣称性能全面超越，标志着两大巨头在该领域的竞争迅速升级。
• 此外，谷歌宣布将Gemini的翻译功能引入谷歌文本翻译，并已推出耳机实时语音翻译的测试版。

相关链接：

• 谷歌紧急上线Gemini Deep Research迎战GPT-5.2，AI“研究代理”大战一触即发
• 谷歌称将Gemini的翻译功能引入谷歌的文本翻译

Google Rushes to Release AI Research Agent, Directly Competing with OpenAI

Brief:

• Ahead of OpenAI's GPT-5.2 release, Google preemptively launched Gemini Deep Research, an AI research agent based on its Gemini 3 Pro model. 🚀
• This tool is specifically designed to handle massive amounts of information and complex research tasks, aiming to improve factual accuracy. It will be made available to developers via a new Interactions API.
• Google plans to integrate Gemini Deep Research into products like Google Search, the Gemini App, and NotebookLM, accelerating its strategic deployment in the era of "agentic AI."
• Following Google's release, OpenAI promptly unveiled GPT-5.2, claiming superior performance across the board, marking a rapid escalation in competition between the two tech giants in this domain. ⚔️
• Additionally, Google announced that it will integrate Gemini's translation capabilities into Google Text Translate and has already launched a beta version of real-time voice translation for headphones. 🎧

Related Links:

• Google Rushes to Launch Gemini Deep Research to Battle GPT-5.2, the AI "Research Agent" War Kicks Off
• Google States It Will Introduce Gemini's Translation Capabilities into Google Text Translate

深圳打造全国首个具身智能示范区，机器人导盲犬已试点上岗

简报：

• 在2025年粤港澳大湾区人工智能与机器人产业大会上，深圳宣布将探索建设全国首个具身智能机器人友好示范城区。
• 该示范区是广东省“1+1+N”具身智能训练体系的一部分，将侧重于机器人在完成封闭训练后，在街区进行开放式训练以激活场景价值。
• 作为应用实例，深圳地铁已在黄木岗交通枢纽试点启用智能导盲犬“小蒜”，该服务在全球轨道交通行业尚属首例，可为视障人士提供自主导航、智能避障等服务。

相关链接：

• 深圳将建全国首个机器人友好示范城区
• 深圳地铁智能导盲犬“小蒜”上线：可自主规划路径、精准导航、视觉识别

Shenzhen to Build China's First Embodied AI Demonstration Zone, Robot Guide Dogs Piloted 🤖

Briefing:

• During the 2025 Guangdong-Hong Kong-Macao Greater Bay Area Artificial Intelligence and Robotics Industry Conference, Shenzhen announced its plan to explore the establishment of China's first robot-friendly demonstration zone for embodied AI.
• This demonstration zone is part of Guangdong Province's "1+1+N" embodied AI training system. It will focus on robots undergoing open-ended training in urban areas after completing closed-environment training, aiming to activate scene-specific value.
• As an application example, Shenzhen Metro has piloted "Xiao Suan," an intelligent guide dog robot, at the Huangmugang transportation hub. This service is a global first in the rail transit industry, offering autonomous navigation and intelligent obstacle avoidance for visually impaired individuals. 🐶✨

Related Links:

• Shenzhen to Build China's First Robot-Friendly Demonstration Zone
• Shenzhen Metro's Intelligent Guide Dog "Xiao Suan" Launched: Capable of Autonomous Path Planning, Precise Navigation, and Visual Recognition