Nathan Lambert@natolambert · 4月30日38

Why push Gemini full gas when all you have to do is walk around the office and pick up billions of dollars off the ground of the Google office.

译为什么要全力推进Gemini，当你只需要在Google办公室里走走，就能从地上捡起数十亿美元。

阿绎 AYi@AYi_AInotes · 4月30日70

Google的AI大模型终于有点像样的的更新更新，这波直接让不少靠AI生成文档的初创公司感受到巨大压力🤯👀 Gemini现在可以在聊天里直接生成Docs、Sheets、Slides、PDF、Word、Excel等主流办公文件， 一句话prompt就能下载， 完全不用复制粘贴、不用手动排版、不用调格式。 支持的范围非常实用：包括带LaTeX公式的学术文档、表格、图表等。 
最震撼的是官方演示——
上传两张手写的潦草笔记，说一句“转成带图表的统计学学习指南PDF”， 几秒钟后，一个标题、目录、公式、表格、视觉元素全都专业排版的完整文档就生成了，直接就能下载。 而且这个功能免费给全球所有Gemini App用户开放，没有Pro门槛，没有地区限制，任何人现在打开手机就能用。 以前的AI，本质上还是聊天机器人：它输出的是文字，你还得自己花时间加工成能交付的东西。 
从这波更新开始，AI输出的不再只是文字，而是可以直接交付的生产力资产——它从“陪你聊天的助手”，升级成了“直接替你干活的员工”。 Google最狠的地方，其实不是模型技术本身，而是把过去二十年积累的Workspace生态，直接变成了Gemini的原生输出底座。 
别人还在卷“谁生成的文字更通顺”，Google直接卷“谁的输出能直接发给老板或客户”。 
那些靠AI生成PPT、PDF、表格吃饭的工具，今天起就从“必备”变成了“可选中间层”，商业模式不得不重新思考。 这才是真正的降维打击。 我理解AI的下一个战场，已经从“谁更会聊天”，彻底转向“谁能直接产出可交付的成果”。 
而Google，借着Workspace的深度集成，先手拉开了一大截领先优势。 对普通用户来说：以后写报告、做PPT、整理笔记，再也不用半夜改格式了，生产力直接起飞。 
对行业来说：这标志着AI从“对话时代”正式进入“交付时代”。

译Google Gemini迎来重磅更新，用户现可在聊天中通过一句话指令，直接生成并下载Docs、Sheets、Slides、PDF等主流办公文件，无需手动复制排版。该功能支持含LaTeX公式的学术文档、表格和图表，且免费向全球Gemini App用户开放。这标志着AI从输出文字的“对话时代”，迈向了直接产出可交付生产力资产的“交付时代”。Google凭借与Workspace生态的深度集成，实现了降维打击，对依赖AI生成文档的初创公司构成巨大压力，并推动行业竞争焦点转向直接产出可用成果。

Chubby♨️@kimmonismus · 4月30日64

Google just quietly proved the AI monetization thesis. Quite interesting earnings! Cloud revenue up 63% to over $ 20b gen AI product revenue growing nearly 800% year-over-year. The backlog nearly doubled to $460 billion in a single quarter. The number of $100M–$1B deals doubled. Multiple billion-dollar-plus contracts signed. But the real story is Search. The prevailing narrative was that AI would cannibalize Google's core business, people get answers from chatbots, stop Googling. The opposite is happening. Search ad revenue grew 19%, queries hit an all-time high. Google turned the biggest existential threat to its business into a growth accelerator.

译谷歌最新财报有力反驳了AI将侵蚀其核心业务的论调。其云收入增长63%至超200亿美元，生成式AI产品收入年增近800%，大额合同储备翻倍。关键转折在于搜索业务：搜索广告收入增长19%，查询量创历史新高。这表明AI非但没有取代传统搜索，反而成为其业务的增长加速器，成功将生存威胁转化为发展动力。

Ethan Mollick@emollick · 4月30日56

Gemini now can create documents, and it is a nice start, but not up to the frontier yet, as you can see from my "LBO of Hogwarts" test. PowerPoints are substantially worse than NotebookLM, spreadsheets are primitive, still no thinking trace, it doesn't think hard enough, either.

译Gemini现在可以创建文档了，这是个不错的开始，但尚未达到前沿水平，正如你从我“霍格沃茨杠杆收购”测试中看到的那样。 PowerPoint比NotebookLM差得多，电子表格功能简陋，仍然没有思考轨迹，它的思考也不够深入。

Sundar Pichai@sundarpichai · 4月30日63

Q1 earnings are in: 2026 is off to a terrific start. Our AI investments and full stack approach are lighting up every part of the business: Search queries are at an all-time high with AI continuing to drive usage. Google Cloud revenue grew 63%, Gemini models have incredible momentum, and it was our strongest quarter ever for consumer AI subs, driven by @GeminiApp. Thanks to our partners + employees around the world. Much more to share on our earnings call in 20 minutes… and at Google I/O in 20 days!

译谷歌2026年第一季度业绩表现强劲，AI投资与全栈策略正全面推动业务增长。公司搜索查询量因AI驱动创下历史新高，Google Cloud收入同比增长63%。Gemini模型发展势头迅猛，以GeminiApp为代表的消费者AI订阅业务也创下季度最佳纪录。公司即将举行财报电话会议，并将在20天后的Google I/O大会上分享更多进展。

Google AI Developers@googleaidevs · 4月30日55

Watch this demo from @thorwebdev to see Gemini 3.1 Flash Live in action as a real-time DJ. The model uses function calling (to the Gemini API) to generate custom 30-second clips using Lyria 3️⃣ Start your own studio session in @GoogleAIStudio: http://goo.gle/3PbcCXJ

译观看 @thorwebdev 的这个演示，看看 Gemini 3.1 Flash 如何作为实时 DJ 实际运作。该模型使用函数调用（调用 Gemini API），通过 Lyria 3️⃣ 生成定制的 30 秒片段。 在 @GoogleAIStudio 中开启你自己的工作室会话：http://goo.gle/3PbcCXJ

Google Gemini@GeminiApp · 4月30日38

This event is happening soon! Join the Gemini Discord here: http://discord.gg/gemini

译这场活动即将开始！在此处加入Gemini Discord：http://discord.gg/gemini [引用 @GeminiApp]：准备好用Gemini Canvas释放你的创造力了吗？🪄 不要错过我们下一次的Discord活动，届时Gemini创意技术专家@DavidMaliglowka将现场演示他最新的Canvas和Nano Banana工作流程，帮助你提升自己的创意提示技巧。 🗓️ 4月29日，星期三 ⏰ 太平洋时间上午11:30 📍 http://discord.gg/gemini

Google AI@GoogleAI · 4月30日52

http://x.com/i/article/2049546144930275328 # The Agentic Era: Unveiling Eighth Generation TPUs A decade in the making, the chips for the agentic era have arrived. At @GoogleCloud's Next '26 event last week, we unveiled our eighth-generation TPUs (the specialized computer chips we build for AI). These chips were specifically designed to handle the two biggest challenges in AI today: training the AI and serving the AI. So… what exactly does that mean? Let’s break it down: TPU 8t: Training the AI Before an AI can help you write an email or plan a trip, it has to "learn" from massive amounts of data. In the past, this could take months of expensive computer time. With TPU 8t, we’ve made that process significantly faster through two key advancements. - More power: It is roughly 3x more powerful than our previous generation of TPUs - More efficiency: We’ve cleared the "traffic jams" that usually slow down AI training. By making data move 10x faster from storage to the chips, we ensure the system is always working at full speed, never sitting idle. - Optimized scaling: In a system this size, parts eventually fail. TPU 8t is designed to automatically detect and reroute around hardware issues at large scale. This ensures that 97% of the resources are spent on productive work, preventing crashes that used to waste days of training time. So now, what used to take months of training now takes only weeks, meaning researchers can experiment and innovate at speed. TPU 8i: Serving the AI (Agents) If the "8t" is for teaching, the 8i is for doing. We built this chip specifically for "AI Agents,” the kind of AI that doesn't just chat with you, but actually acts for you (ex: booking a flight, managing a calendar, etc). To take action, an AI needs to "think" and "reason" through multiple steps very quickly, which TPU 8i enables through these advancements: - Better thinking: We tripled the chip’s internal memory so it can handle more complex logic. - More cost effective: It offers 80% better performance for every dollar spent. For a business, that means you can help twice as many customers without increasing your tech budget. - Latency: At the chip level, we have integrated a new engine which reduces latency by an additional 5x. Powering the Next Decade Whether it's a scientist training a new medical model or a business getting some much needed customer support help, these chips provide the raw power needed to make that future a reality.

译在Google Cloud Next '26大会上，谷歌正式推出专为智能体时代设计的第八代TPU芯片，分别针对AI训练与服务两大核心挑战。TPU 8t专注于训练，其性能约为前代的3倍，并通过加速数据移动和优化硬件容错，将原本需数月的训练时间缩短至数周。TPU 8i则专为执行复杂任务的AI智能体服务，内存扩大三倍以支持多步推理，每美元性能提升80%，延迟降低5倍，助力企业以更低成本扩展服务规模。这些芯片将为医疗研究、客户支持等广泛场景提供核心算力，推动AI应用创新。

TestingCatalog News 🗞@testingcatalog · 4月30日68

GOOGLE 🚨: Gemini now can generate Docs, Sheets, Slides, and PDFs directly in the chat. Available to all users already 👀

译GOOGLE 🚨：Gemini 现在可以直接在聊天中生成文档、表格、幻灯片和 PDF。 已面向所有用户开放 👀 [引用 @joshwoodward]：Gemini 新功能：生成文件并导出 告诉 Gemini 你想创建什么内容和格式，它现在就能为你完成工作。 目前支持： 📄 Google 文档、Word (.docx) 和 PDF 📊 Google 表格、Excel (.xlsx) 和 CSV 🖥️ Google 幻灯片 🛠️ Markdown、LaTeX、TXT、RTF 现已全球全面上线！

Josh Woodward@joshwoodward · 4月30日68

New in Gemini: Generate files and export them Tell Gemini what you want to create and the format, and it now does the work for you. Now supporting: 📄 Google Docs, Word (.docx) & PDFs 📊 Google Sheets, Excel (.xlsx) & CSV 🖥️ Google Slides 🛠️ Markdown, LaTeX, TXT, RTF Available now on all surfaces globally!

译Gemini 新功能：生成文件并导出 告诉 Gemini 你想创建什么以及格式，它现在就能为你完成。 现已支持： 📄 Google 文档、Word (.docx) 和 PDF 📊 Google 表格、Excel (.xlsx) 和 CSV 🖥️ Google 幻灯片 🛠️ Markdown、LaTeX、TXT、RTF 现已面向全球所有平台推出！

Sundar Pichai@sundarpichai · 4月30日67

You can now ask Gemini to create Docs, Sheets, Slides, PDFs, and more directly in your chat. No more copying, pasting, or reformatting, just prompt and download. Available globally for all @GeminiApp users.

译你现在可以直接在聊天中让Gemini创建Docs、Sheets、Slides、PDF等文件。无需再复制、粘贴或重新格式化，只需输入指令并下载即可。 此功能已面向全球所有@GeminiApp用户开放。

Google Gemini@GeminiApp · 4月30日60

You can now generate a variety of downloadable files, including PDFs, @GoogleWorkspace files, Microsoft Word & Excel, and more directly in your chats with Gemini. Tell Gemini what content to create and the file format you want when you prompt without having to upload a template.

译现在您可以在与Gemini的聊天中直接生成多种可下载文件，包括PDF、@GoogleWorkspace文件、Microsoft Word & Excel等。 只需在提示时告诉Gemini要创建的内容和所需文件格式，无需上传模板。

TestingCatalog News 🗞@testingcatalog · 4月29日36

Google is working on Mind Map customization for NotebookLM and a new integration with Google Play Books. What's Coming? 👀 > Users will be able to instruct NotebookLM to build a Mind Map for a specific topic or a set of sources. > Users will be able to use Google Play Books as sources. "Turn bestsellers into personalized insights. Add full-length books from leading authors to your notebooks."

译Google正在为NotebookLM开发思维导图定制功能以及新的Google Play Books集成。 即将推出什么？👀 > 用户将能指导NotebookLM为特定主题或一组资料构建思维导图。 > 用户将能把Google Play Books用作资料来源。 "将畅销书转化为个性化见解。将知名作者的全本著作添加到你的笔记本中。"

Demis Hassabis@demishassabis · 4月29日60

Excited to collaborate with the Korea Ministry of Science and ICT (@msitmedia) to use AI to accelerate scientific discovery and to invest in Korea’s next generation of talent. Many thanks for hosting us @msitminister - look forward to working together!

译Google DeepMind首席执行官Demis Hassabis与韩国科学技术信息通信部（MSIT）签署谅解备忘录，合作利用AI加速科学发现并投资韩国下一代人才。此次合作在AlphaGo问世十年后举行，标志着AI发展的新转折点。双方将聚焦三大核心领域：科学技术研究协作、AI人才培养以及AI安全治理。强调AI发展需全球研究能力与产业基础联动，无法单靠一国或一企完成。AlphaFold等案例已证明AI能变革科学发现速度，未来十年将是把AI潜力转化为现实的关键期。

Rohan Paul@rohanpaul_ai · 4月29日58

Google is turning consultants into its AI delivery network with a $ 750M fund for firms like McKinsey, Accenture, and Deloitte to help companies build and scale agentic AI. Consulting firms need this because classic consulting work, such as research, slide drafting, process mapping, and software planning, is exactly the kind of work AI systems are starting to automate. AI startups need consultants because big companies rarely buy new tools just because the model is powerful, since they need someone to connect it with data, workflows, security rules, and staff habits. Agentic AI means software that does not only answer questions, but can plan steps, call tools, move through business systems, and complete tasks with less human steering. So Google’s bet is that McKinsey can find the business problem, Google can provide the AI stack, and the client can turn a pilot into a working system across teams. OpenAI’s reported push to sell Codex through Accenture, Capgemini, and PwC points to the same shift, where AI coding tools become enterprise software only after consultants package them into training, governance, and rollout plans. --- businessinsider. com/consulting-mckinsey-accenture-bcg-ai-silicon-valley-enterprise-partnerships-2026-4

译谷歌正通过设立7.5亿美元基金，将麦肯锡、埃森哲等顶级咨询公司转变为自己的“AI交付网络”，以帮助企业构建和规模化智能体AI。其核心逻辑在于，咨询公司的传统工作正被AI自动化，而它们擅长连接新技术与企业数据、工作流及安全规则，成为AI落地的关键桥梁。谷歌的布局是：咨询公司发现业务问题，谷歌提供AI技术栈，客户则将试点推广至全公司。OpenAI通过埃森哲等渠道销售Codex的举措，也印证了同一趋势——AI工具需经咨询公司包装成包含培训与治理的解决方案，才能成为真正的企业软件。

ginobefun@hongming731 · 4月29日57

玩转 Gemini 3.1 TTS：音频标签与提示词技巧指南

译Google AI推出的Gemini 3.1 TTS模型新增音频标签功能，开发者可通过方括号内的标签直观控制语音风格、语速和表达。关键使用技巧包括：标签需用方括号包裹并置于期望转换点，避免直接相邻；使用[slow]、[fast]控制语速，[short pause]制造戏剧停顿；还能通过[cackles]、[whispers]等标签精细操控发声。这些提示词技巧适用于构建语言学习工具、互动播客应用或自适应客服等多种场景，赋能开发者高效利用模型进行音频创作。

Berryxia.AI@berryxia · 4月29日58

Google Gemma 官方教你本地跑 Coding Agent！ 本地完美组合来了： • Pi Agent • Gemma 4 26B 模型 • LM Studio / Ollama / llama.cpp 等 serving engine 完全离线运行、零 API 费用、100% 隐私保护、零延迟！本地开发者 Agentic 开发神器！ 附 @patloeber 详细一步步搭建教程👇 https://patloeber.com/gemma-4-pi-agent/

Jeff Dean@JeffDean · 4月29日48

Google Translate is turning 20! 🎉. There are 20 fun facts and tips in the thread below. Translate is one of my favorite Google products because it brings us all closer together! I've been involved with a couple of things over the years. The first was our deployment of the initial system in 2006, which provided a huge leap forward in quality because it used a much larger 5-gram language model trained on trillions of words of text (indeed, probably the first trillion token language model training in the world: paper has some nice heads showing scaling-law-like quality improvement from scaling to more data/compute). See "Large Language Models in Machine Translation", Thorsten Brants, Ashok C. Popat, Peng Xu, Franz J. Och and Jeffrey Dean, https://aclanthology.org/D07-1090/ The second major collaboration was in 2016 when we moved Translate over from a statistical machine translation approach to using deep neural networks.  This approach relied on two key innovations.  The first was Google's work on Sequence-to-Sequence models (https://arxiv.org/abs/1409.3215).  The second was our development of TPUs, custom cups that improved the performance of inference for deep neural networks by 30-80X over existing CPUs and GPUs of the day (and reduced latency by 15-30X).  This made launching compute-intensive language model services like Translate feasible for hundreds of millions of users. See "In-Datacenter Performance Analysis of a Tensor Processing Unit",  Norman P. Jouppi et al.  https://arxiv.org/abs/1704.04760 GNMT paper: "Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation",  Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V. Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, Jeff Klingner, Apurva Shah, Melvin Johnson, Xiaobing Liu, Łukasz Kaiser, Stephan Gouws, Yoshikiyo Kato, Taku Kudo, Hideto Kazawa, Keith Stevens, George Kurian, Nishant Patil, Wei Wang, Cliff Young, Jason Smith, Jason Riesa, Alex Rudnick, Oriol Vinyals, Greg Corrado, Macduff Hughes, and Jeffrey Dean, https://arxiv.org/abs/1609.08144 Most recently, we have advanced Translate further using Gemini models. Each of these advances relied on research that have major quality leaps over the existing status quo translation approaches, bringing better quality and connectedness to all of our Translate users! 🎉

译Google Translate迎来20周年，其发展依赖多次技术飞跃。2006年部署基于万亿词训练的5-gram语言模型，实现质量突破；2016年转向深度神经网络，结合Sequence-to-Sequence模型和TPUs，性能提升30-80倍、延迟降低15-30倍，使大规模服务成为可能；近期集成Gemini模型进一步优化。这些进步均基于前沿研究，每次都为翻译质量带来显著提升。作为Google机器学习工作的初始实验，Google Translate最常见翻译短语如“thank you”体现了其连接全球用户的使命。

Rohan Paul@rohanpaul_ai · 4月29日62

Bloomberg: Google decided to back away from a $ 100M Pentagon drone swarm contest after first making the cut, exposing how divided Big Tech still is over military AI. The project aimed to turn voice commands like directional orders into machine instructions for groups of autonomous drones. Google’s exit appears less about raw capability than about internal limits on what kind of defense work the company is willing to own. --- bloomberg. com/news/articles/2026-04-28/google-drops-out-of-pentagon-drone-swarm-contest-after-advancing

译彭博社报道，谷歌在入围后决定退出美国国防部一项价值1亿美元的无人机集群竞赛。该项目旨在将语音指令转化为对自主无人机群的机器指令。谷歌的退出并非由于技术能力不足，而更多源于公司内部对愿意承担的国防工作类型设定了限制。这一事件凸显了大型科技公司在军事人工智能应用上仍然存在深刻分歧。

Chubby♨️@kimmonismus · 4月29日46

SandboxAQ spun out of Google, raised $950M+, and is backed by NVIDIA. Everyone is talking about LLMs. Almost nobody is talking about LQMs. Sandbox bet: Large Quantitative Models that simulate physics and chemistry to invent new drugs and materials. I talked to their GM of AI Simulation Nadia Harhen about why LQMs might matter more than LLMs for the physical world. Our newest (and second) Podcast-Episode of Superintelligence Podcast - out now! Link in comments

译SandboxAQ从谷歌分拆出来，筹集了超过9.5亿美元，并获得了英伟达的支持。 每个人都在谈论LLMs。 几乎没有人谈论LQMs。 Sandbox的赌注：通过模拟物理和化学的大型定量模型来发明新药物和新材料。 我与他们的AI模拟总经理Nadia Harhen讨论了为什么对于物理世界来说，LQMs可能比LLMs更重要。 我们最新（也是第二期）的《超级智能播客》节目现已发布！链接在评论中

Sundar Pichai@sundarpichai · 4月29日46

Hello. How are you? Thank you. I love you. Please. Some of the most frequently translated phrases of the past 20 years! Google Translate began twenty years ago with a mission to help people understand one another, regardless of the language they speak. What started as a small experiment has become a global tool that helps over 1 billion users every month. In that time Translate has evolved from simple pattern matching to true understanding. In 2006, it relied on statistical machine learning to look for patterns in small word clusters. By 2016, we pioneered a shift to neural networks to move beyond literal word-for-word translations, and today we’re using our powerful Gemini models to make Translate even more helpful. We are moving from text to fluid, real-time conversations. With our latest models, you can even use your headphones as a personal interpreter that preserves your original tone and cadence - it’s an amazing experience! One of the interesting things about AI is that as we make progress, we begin to take it for granted. If you met a person who could translate across a hundred languages faster than any human can, you would be so impressed. Today, one product does that for nearly 250 languages, and we kind of just shrug. Being able to say thank you in 250 languages is not something I take for granted. So to the 1 billion who use Google Translate - merci, dhanyavaad, arigatō, gracias, and thank you! Let’s see what the next 20 years will bring.

译谷歌翻译迎来二十周年，已从最初的简单模式匹配发展为每月服务超10亿用户的全球工具。其技术历经三个阶段：2006年依靠统计机器学习分析词簇，2016年转向神经网络实现超越字面的翻译，如今借助Gemini模型进一步提升能力。当前发展重点正从文本翻译转向流畅的实时对话，最新模型甚至能通过耳机充当口译器，并保留用户原有的语调和节奏。尽管AI翻译已支持近250种语言，人们却逐渐将其视为常态。谷歌对此表达感谢，并展望未来二十年的技术突破。

Chubby♨️@kimmonismus · 4月28日70

Google just signed a deal letting the Pentagon use its AI models for classified work and "any lawful government purpose." This comes despite over 600 employees urging CEO Sundar Pichai to reject the agreement, and marks a dramatic reversal from 2018 when Google pulled out of Project Maven after employee backlash. Google now joins xAI and OpenAI in having classified Pentagon AI deals, with terms that appear even more permissive than OpenAI's. The contract includes language saying Google's AI "is not intended for" mass surveillance or autonomous weapons without human oversight, but legal experts say this wording is not legally binding. Notably, the deal also requires Google to adjust its AI safety filters at the government's request. This all follows Anthropic's public refusal to drop its red lines on those exact use cases, which led to the Pentagon declaring Anthropic a supply chain risk, a designation Anthropic is currently fighting in court.

译谷歌已与五角大楼签署协议，允许其AI模型用于机密工作及“任何合法的政府目的”，此举无视了超600名员工的反对，并逆转了其2018年因员工抗议退出Project Maven的立场。协议条款看似比OpenAI的同类合约更为宽松，虽声明AI“不拟用于”大规模监控或无人监督的自主武器，但法律专家指出该措辞缺乏约束力。协议还要求谷歌应政府要求调整AI安全过滤器。这与Anthropic因拒绝在类似用途上妥协而被五角大楼列为供应链风险形成对比。

TestingCatalog News 🗞@testingcatalog · 4月28日49

ICYMI: Gemini can now generate Docs and Sheets on web and mobile. Not sure when it was added though. Slides are not working for now but looking at Gemini for Business, we will likely get them too, as well as an inline editor potentially.

译你可能错过了：Gemini 现在可以在网页和移动端生成 Docs 和 Sheets。不过不确定这个功能是何时添加的。 目前 Slides 还不能用，但考虑到 Gemini for Business，我们很可能也会获得该功能，或许还会有一个内联编辑器。

Berryxia.AI@berryxia · 4月28日64

一个完全本地的 Agent，就生活在你的浏览器里。 由 Gemma 4 E2B 和 WebGPU 驱动，它使用原生工具调用来实现： 🔍 搜索浏览历史 📄 阅读并总结页面内容 🔗 管理标签页 100% 本地运行！无需任何服务器！

Google Gemini@GeminiApp · 4月28日31

Ready to unlock your creativity with Gemini Canvas? 🪄 Don’t miss our next Discord event to see Gemini Creative Technologist @DavidMaliglowka live demo his latest Canvas and Nano Banana workflows to help you advance your own creative prompting techniques. 🗓️ Wednesday, April 29th ⏰ 11:30 AM PT 📍 http://discord.gg/gemini

译准备好通过Gemini Canvas释放你的创造力了吗？🪄 别错过我们下一次Discord活动，届时Gemini创意技术专家@DavidMaliglowka将现场演示他最新的Canvas和Nano Banana工作流程，帮助你提升创意提示技巧。 🗓️ 4月29日星期三 ⏰ 太平洋时间上午11:30 📍 http://discord.gg/gemini

Google AI Developers@googleaidevs · 4月28日52

Zoom in on how @GoogleGemma 4 is optimized to handle high-concurrency serving for complex tasks (such as generating SVGs) — on a single GPU. ✓ 10+ sessions are sent to the 26B A4B model ✓ The system routes, accelerates, and processes those workloads — without bottlenecking ✓ A live dashboard visually tracks the load balancing in real time, displaying active slots, context sizes, and token generation speeds Watch the demo to see it in action ⬇️

译深入了解 @GoogleGemma 4 如何优化以在单个 GPU 上处理高并发复杂任务（例如生成 SVG）。 ✓ 10 多个会话被发送到 26B A4B 模型 ✓ 系统路由、加速并处理这些工作负载——没有瓶颈 ✓ 实时仪表板可视化跟踪负载均衡，显示活动槽位、上下文大小和令牌生成速度 观看演示视频以了解实际运行情况 ⬇️

Jeff Dean@JeffDean · 4月27日56

The video of my conversation with Amin Vahdat, @gilbert, and @djrosent at Cloud Next last week is now up. https://youtu.be/BpnJYJmbXcM?si=vUY3hI_aDX8K6gco Thanks for a great conversation!

译视频记录了在Cloud Next大会上与Amin Vahdat及AcquiredFM主持人的对话，核心围绕谷歌最新发布的TPU v8t和v8i芯片展开讨论。对话内容基于官方博客公布的芯片细节，探讨了其在“智能体时代”的基础设施意义与技术亮点。

Chubby♨️@kimmonismus · 4月27日68

Google just broke a decade-long tradition. At Cloud Next 2026, the company unveiled not one, but two new AI chips, the TPU 8t for training and TPU 8i for inference. For the first time ever, Google is splitting its custom silicon into specialized architectures instead of relying on a one-size-fits-all design. The TPU 8t superpod packs 9,600 liquid-cooled chips delivering 121 FP4 ExaFlops of peak compute, roughly a 3x leap over the previous generation. The TPU 8i delivers 80% better performance-per-dollar than its predecessor, with triple the on-chip memory and a new Boardfly topology that cuts network latency in half. The important aspect: Anthropic, Meta, and now OpenAI are buying multi-gigawatt allocations of TPU capacity. OpenAI booking Google silicon is a first visible crack in NVIDIA's grip on frontier AI training. Broadcom co-designed the TPU 8t, while MediaTek handles the TPU 8i, both fabbed by TSMC. NVIDIA still holds 81% of the AI chip market, but the era of serious competition has officially begun.

译Google在Cloud Next 2026上首次将定制芯片拆分为专用架构，推出训练芯片TPU 8t与推理芯片TPU 8i。TPU 8t超级模块配备9600个液冷芯片，峰值算力达121 FP4 ExaFlops，较前代提升约3倍；TPU 8i的性价比提升80%，片上内存增至三倍，并通过新拓扑结构将网络延迟减半。Anthropic、Meta及OpenAI均已采购千兆瓦级TPU算力，其中OpenAI首次采用Google芯片，动摇了NVIDIA在前沿AI训练市场的垄断地位。两款芯片分别由Broadcom和MediaTek共同设计，TSMC代工。尽管NVIDIA仍占据81%的AI芯片市场份额，但实质性的竞争时代已拉开序幕。

Google DeepMind@GoogleDeepMind · 4月27日37

A decade ago in Korea, AlphaGo showed AI’s potential. Together with the Korean government, we’re now looking at how this technology can help accelerate scientific discovery and create new opportunities for economic growth across the region. 🇰🇷 Find out more → https://goo.gle/4wbqUbJ

译十年前在韩国，AlphaGo展现了AI的潜力。 如今我们与韩国政府共同探索这项技术如何加速科学发现，并为整个区域创造经济增长的新机遇。🇰🇷 了解更多 → https://goo.gle/4wbqUbJ

Chubby♨️@kimmonismus · 4月27日63

Google's TPU v8 and Huawei's Ascend NPU platform: the global Chipwar just began At Cloud Next 2026, Google unveiled its eighth-generation TPU as two separate chips for the first time: the TPU 8t for training and the TPU 8i for inference, claiming up to 2.8x faster training and 80% higher performance per dollar for inference compared to last year's Ironwood. The 8t was designed by Broadcom, the 8i by MediaTek, applying mobile-edge efficiency logic to inference while maximizing raw throughput on training. The 8t connects up to 9,600 accelerators via optical-circuit switches, dwarfing NVIDIA's 576-GPU NVLink domain, and a new Virgo network fabric scales beyond one million chips for a single training job. Google is also replacing x86 hosts with its own Arm-based Axion CPUs, completing full vertical control from host to accelerator to network. The message is clear: the general-purpose AI accelerator is a fading category. DeepSeek V4 on Huawei Ascend: China's parallel infrastructure takes shape DeepSeek's V4 release is the more geopolitically consequential event. The 1.6 trillion-parameter V4-Pro is the first major frontier model to validate both training and inference on Huawei's Ascend NPU platform alongside NVIDIA GPUs. The nuance: DeepSeek adapted only part of V4's training for Chinese chips and confirmed Ascend for inference, while pre-training of V4-Pro likely still relied on NVIDIA silicon. Is this a novum? Yes. No frontier-class model has ever publicly validated on non-NVIDIA hardware at this scale. More importantly, DeepSeek is tying future pricing to Huawei's Ascend 950 production ramp in H2 2026, making this an economic bet, not a symbolic gesture. V4-Pro costs $3.48 per million output tokens versus $30 for GPT-5.4 and $25 for Claude Opus 4.6. The real story isn't whether V4 beats Western models on benchmarks (it doesn't quite), but whether the hardware decoupling U.S. sanctions were designed to prevent is now irreversibly underway.

译谷歌在Cloud Next 2026上首次将TPU v8拆分为训练芯片TPU 8t和推理芯片TPU 8i，宣称训练速度提升2.8倍，推理性价比提高80%，并通过自研Arm架构Axion CPU实现全栈垂直控制。同时，DeepSeek V4-Pro成为首个在华为昇腾NPU平台上完成训练与推理验证的前沿大模型，其定价与昇腾950芯片量产计划挂钩，输出成本远低于主流西方模型。这标志着美国制裁试图阻止的硬件脱钩可能已不可逆转，全球AI芯片竞争进入新阶段。

Berryxia.AI@berryxia · 4月27日55

小耳做的这个小插件，解决了右键批量命名的问题。 Apple macOS自带的批量修改的问题是只能是批量改一样前缀和后缀的名字，不够智能和方便。 这个就是纯使用AI多模态的Gemini模型来给你识别，可能要消耗token，其实可以直接使用Gemma 4或者qwen多模态本地使用小模型可能成本更低更方便。 推荐大家体验下载使用👇🏻

译一款名为“小耳”的开发者制作的macOS右键工具，利用AI多模态模型（如Gemini）智能识别文件内容，并自动将其重命名为“内容+日期”的格式，解决了系统自带批量重命名功能不够智能的问题。该工具以Quick Action形式集成，无需安装新应用或后台进程，支持图片、视频、PDF、Word等多种文件类型的批量处理。其关键优势包括处理在本地完成、可使用Gemini Flash免费额度、支持撤销操作，并建议用户也可考虑使用Gemma或Qwen等本地小模型以降低成本。

Chubby♨️@kimmonismus · 4月27日36

This years WWDC will be the most exciting one. - new Apple CEO - first time a useful AI model deeply integrated into iOS (Gemini) - excited for new macOS features. I wish I would be there in person

译今年的WWDC将是最令人兴奋的一届。 - 新任苹果首席执行官 - 首次将实用的AI模型深度集成到iOS中（Gemini） - 期待全新的macOS功能。 真希望我能亲临现场

TestingCatalog News 🗞@testingcatalog · 4月26日32

GOOGLE I/👀: Google is working on Website generation for its Pomelli experiment. > Considering that Pomelli is a Marketing AI agent, a new Websites feature may help users generate and potentially host landing pages. > A new Catalog feature is also in the works that may help SMBs upload catalog items in batches and generate marketing assets in bulk.

译谷歌I/👀：Google 正在为其 Pomelli 实验开发网站生成功能。 > 考虑到 Pomelli 是一个营销 AI 助手，新的网站功能可能帮助用户生成并托管落地页。 > 新的目录功能也在开发中，或可帮助中小企业批量上传目录商品并批量生成营销素材。

TestingCatalog News 🗞@testingcatalog · 4月25日33

Google is working on a "Usage Limits" section for Gemini and a new "Images" tab. > At this point, it is unclear if we should expect a new Image model to arrive during Google I/O or if we would see a new Images section with extra features for image editing and more. > Usage Limits tab aligns with a broader direction for Gemini, where we would expect its Desktop app to be expanded to include more agentic features. AI Studio already shares Usage Limits with Gemini. Super Gemini App 👀

译Google正在为Gemini开发"使用限制"板块和新的"图像"标签。 > 目前尚不清楚我们是否能在Google I/O期间迎来新的图像模型，或者是否会看到带有图像编辑等额外功能的新图像板块。 > 使用限制标签符合Gemini更广泛的发展方向，我们预计其桌面应用将扩展包含更多智能体功能。AI Studio已与Gemini共享使用限制。 超级Gemini应用 👀

Chubby♨️@kimmonismus · 4月25日39

By soon I expect them to mean may 18th-soon. Google I/o will probably have some nice surprises for us

译我预计他们说的“很快”是指5月18日前后。Google I/O大会可能会给我们带来一些惊喜

Chubby♨️@kimmonismus · 4月24日36

I assume we will see a big release on google i/o on may 18th. again: google doesnt have the compute constraint most frontier labs have. Expect a high jump on evals and usage soon.

译我猜我们会在5月18日的谷歌I/O大会上看到重大发布。 再次强调：谷歌没有大多数前沿实验室面临的计算资源限制。 预计评估分数和使用量很快会有大幅提升。

Sundar Pichai@sundarpichai · 4月23日

TPU 8t, optimized for training and TPU 8i, optimized for inference. Looking good!

译TPU 8t 针对训练优化，TPU 8i 针对推理优化。 看起来不错！

Rohan Paul@rohanpaul_ai · 4月22日

AI demand is growing fast. Google Cloud now processes 16 billion+ tokens per minute via direct API use by their customers, up from 10 billion last quarter.

译AI 需求快速增长。 Google Cloud 目前通过客户直接调用 API，每分钟处理 16 billion+ tokens，而上季度为 10 billion。

Google Gemini@GeminiApp · 4月22日

If you’re not already using Gems to optimize your workflow in Gemini, it’s time to start. Gems allow you to quickly reuse a prompt and add reference files. Open the side panel, create a gem, and turn your repetitive tasks into a single click.

译如果你还没有使用 Gems 来优化 Gemini 中的工作流程，现在是时候开始了。 Gems 让你能够快速复用提示词并添加参考文件。 打开侧边面板，创建一个 gem，将重复性任务变成一键操作。

TestingCatalog News 🗞@testingcatalog · 4月22日34

GOOGLE 🚨: REFERENCES TO AN UPDATED DEEP RESEARCH AND DEEP RESEARCH MAX MODELS HAVE BEEN SPOTTED! - deep-research-max-preview-04-2026 - deep-research-preview-04-2026 Google Deep Max Ultra Pro 👀

译GOOGLE 🚨: 已发现关于更新版深度研究和深度研究MAX模型的引用！ - deep-research-max-preview-04-2026 - deep-research-preview-04-2026 Google Deep Max Ultra Pro 👀