跳到正文
9月8日周二
11月10日周二
  1. AI at Meta65

    Meta 宣布将 Muse Ember 连同 Muse Wonder 和 Muse Spark 1.3 一起引入 Google Cloud。被引用的发布内容称,Muse Ember 是 Meta 最小、最高效的模型,面向复杂推理执行和日常智能体工作,可跨应用承接工作流、按顺序调用工具、运行测试并做针对性代码修复,具备 1M-token 上下文窗口和文本、图像、音频、视频原生理解能力。

    引用AI at Meta@AIatMeta

    Today we're introducing Muse Ember, our smallest and most efficient model, built for complex reasoning execution and everyday agent work. Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.

    推荐理由:官方公布 Muse Ember 等模型接入 Google Cloud,读者可据此了解这些模型在云平台上的可用范围。

10月29日周四
  1. The Rundown AI65

    Google 发布 Gemini 4 Carbon,登顶基准榜但暂未公开

    Google 发布面向网络安全和长时间编码任务的前沿模型 Gemini 4 Carbon,在 Google 图表 17 项基准中登顶 15 项,但暂未公开,仅部分网络安全防御人员获得访问权限。

    推荐理由:原文给出 Carbon 在 17 项基准中 15 项领先及与 Opus 5.5、Astra 的分数对比,可据此判断其位置与尚未开放的限制。

  2. meng shao65

    Google Research 发布 SLEEP 框架,让 LLM 在清醒交互与睡眠巩固之间周期切换,以减轻灾难性遗忘。睡眠阶段会解锁预留的 MoE 记忆块,用 GKD 与 LTI 把短时上下文写入长时参数,再通过随机激活专家生成合成数据、梯度筛选后做监督微调。

    引用Google Research@GoogleResearch

    Today on the blog, learn about our new instance of a “sleep” phase in continual learning, which enables LLMs to continuously integrate new facts and enhance reasoning while mitigating catastrophic forgetting or static knowledge cutoffs. More →https://goo.gle/4iICkyD

    推荐理由:原文梳理了 SLEEP 在清醒与睡眠间切换的机制,并列出翻译、长上下文与数学推理的对比结果。

10月28日周三
  1. The Rundown AI65

    Google DeepMind 发布 Genie 4 世界模型,支持原生音频与可对话角色

    Google DeepMind 发布新一代世界模型 Genie 4,可将提示词实时转化为可探索的逼真场景,并首次由模型自行生成背景音与音乐,场景中的角色也能进行实时对话。

    推荐理由:Genie 4 在实时生成场景之外加入原生音频与可对话角色,读者可据此判断世界模型正从漫游工具转向可编排的创作工具。

10月21日周三
  1. OpenAI News62

    OpenAI 模型将上线 Google Cloud 的 Gemini Model Garden

    OpenAI 与 Google Cloud 宣布,OpenAI 的前沿模型和 Codex 工作流将接入 Gemini 的 Model Garden,并通过 Google Cloud Marketplace 提供。原文未披露具体模型版本、上线时间和定价。

    推荐理由:OpenAI 模型与 Codex 工作流接入 Google Cloud 的 Gemini Model Garden,读者可了解企业采购与部署路径的变化。

11月19日周四
11月13日周五
  1. 🚨 AI News | TestingCatalog88

    Google 正式发布 Gemini 4 Hydrogen,官方称其完成任务速度比 3.8 Flash 快 2.5 倍,并提供 4 档思考深度,其中新增 Extra 档。该模型今日起向 Google AI Studio、Gemini API、Gemini App、Search 及 Google AI Pro 和 Ultra 订阅用户推送;作者引用的对比图显示,它在 SWE-sweep、Agents' Last Exam v1.1、AutomationBench-AA 等基准上高于 Claude Sonnet 5.5 和 GPT-6.1 Terra,但在 Terminal-Bench 4.1、OSWorld-2.1 等项上落后。

    引用Sundar Pichai@sundarpichai

    Following Carbon + Argon, Gemini 4 Hydrogen brings cutting-edge intelligence to everyone. Built for speed and efficiency for everyday tasks, coding and agentic workflows. Compared to 3.8 Flash, it finishes tasks 2.5x faster. Also has 4 thinking levels, so you only wait and pay for the reasoning depth a task requires. Rolling out today to all tiers of @GoogleAIStudio, the Gemini API, @GeminiApp, Search and paid Google AI Pro and Ultra subscribers in Google @Antigravity.

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Hydrogen 模型》

  2. 宝玉88

    Google 于 11 月 12 日发布 Gemini 4 Hydrogen,接替 Gemini 3.8 Flash,主打快和省,对标 Anthropic 的 Claude Sonnet 5.5 和 OpenAI 的 GPT-6.1 Terra。

    引用Sundar Pichai@sundarpichai

    Following Carbon + Argon, Gemini 4 Hydrogen brings cutting-edge intelligence to everyone. Built for speed and efficiency for everyday tasks, coding and agentic workflows. Compared to 3.8 Flash, it finishes tasks 2.5x faster. Also has 4 thinking levels, so you only wait and pay for the reasoning depth a task requires. Rolling out today to all tiers of @GoogleAIStudio, the Gemini API, @GeminiApp, Search and paid Google AI Pro and Ultra subscribers in Google @Antigravity.

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Hydrogen 模型》

  3. Demis Hassabis88

    Google 发布 Gemini 4 Hydrogen,主打日常编码、智能体任务和通用场景下的速度与智能。该模型面向 Google AI Studio、Gemini API、Gemini 企业版、Gemini 应用、Google Search 各档位开放,Google AI Pro 和 Ultra 付费订阅用户可在 Google Antigravity 中使用。官方基准图显示,其在 SWE-sweep 上为 13.5%,Agents' Last Exam v1.1 为 71.6%,HLE-Diamond 为 58.8%。

    引用Logan Kilpatrick@OfficialLoganK

    Introducing Gemini 4 Hydrogen, an extremely capable and fast model built to bring Gemini 4 to everyone! Available to all tiers of Google AI Studio, Gemini API, Gemini for business, the Gemini app, Google Search, and paid Google AI Pro + Ultra subscribers in Google Antigravity.

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Hydrogen 模型》

11月11日周三
  1. 宝玉88

    Google 于 11 月 10 日将 Gemini 4 Carbon 开放给付费开发者、企业和 Google AI Ultra 订阅用户,此前 10 月 28 日仅限通过审查的网络安全防御团队使用。

    引用Sundar Pichai@sundarpichai

    Gemini 4 Carbon is rolling out today. It’s built for sustained, long-horizon tasks, reasoning deeply through complex software engineering, cybersecurity, and scientific discovery. Coming globally via the Gemini API in @GoogleAIStudio and Gemini for business, and for Ultra subscribers in @antigravity, the @GeminiApp, and AI Mode in Search.

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Carbon 模型》

  2. Rohan Paul88

    Google DeepMind 团队用 AlphaProof Code 框架对 Signal 协议底层机器码完成形式化验证,证明其关键安全属性成立。该框架由人类引导编写 Lean 规范、AArch64 机器码语义和自主证明智能体组成,约 6.5M 行机器可检验的 Lean 证明代码建立在约 26,000 行可信规范与语义之上。

    引用Pushmeet Kohli@pushmeet

    Announcing a major milestone achieved by my team @GoogleDeepMind: we have proven key security properties of the machine code underlying the Signal messaging protocol. This is the first time the actual production code of a widely deployed protocol has been formally verified – by any method, with or without AI – and at this depth. More than a research breakthrough, it’s a prime example of how frontier AI can be a force multiplier to counter the speed and scale of modern cyber threats. Formal verification offers a higher standard of security; however, it’s extremely complex and time consuming, which is why in many cases it’s considered to be impractical. Similar efforts could take teams decades of combined effort. We did it in a month’s time. Using AlphaProof Code, Google DeepMind's new agentic framework for formal verification that runs on LLMs including Gemini Carbon, we verified the Signal protocol down to the actual code running on devices.  This milestone lays the groundwork for a future in which vital systems, from medical device firmware to power grid software, ship with mathematical proof that their protections work. Reaching that future will take continued research from our community. In that spirit, we're open-sourcing our formal specifications and verification artifacts so others can build on this work. On a personal note, this moment is of deep significance to me. I started my own research journey more than two decades ago as an intern working on checking the consistency of complex software systems. Seeing this grand challenge made practical with modern AI is extraordinary. I'm extremely proud of my team members: Swarat Chaudhuri, Andrew Ferraiuolo, Anton Kovsharov, Tej Chajed, Sam Schlesinger,  Karn Seth, Jonathan Katz, and Charles Sutton for taking on this grand challenge and delivering, and grateful to the Signal team for their partnership. https://deepmind.google/blog/formal-verification-signal-protocol

    同一件事,精选收录的是《Google DeepMind 推出 AlphaProof Code,形式化验证 Signal 协议代码》

10月29日周四
11月19日周四
  1. Rohan Paul85

    Google 发布新模型 Gemini 4 Neon,在 Artificial Analysis 智能指数上得 61 分,排名第三,定价为每 100 万输入/输出 token 3/15 美元。

    引用Sundar Pichai@sundarpichai

    Gemini 4 Neon: frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows. Begins rolling out today in the Gemini API in @GoogleAIStudio, Gemini for business, for paid Google AI Pro and Ultra subscribers in Google @Antigravity, and for Google AI Pro and Ultra subscribers in the @GeminiApp.

    同一件事,精选收录的是《Google 发布 Gemini 4,上线 Gemini App、Antigravity 与 API》

11月11日周三
10月29日周四
10月28日周三
11月19日周四
  1. Dongxi 东锡 NLP82

    Google 发布 Gemini 4 Neon,称其具备前沿智能与快速任务完成能力,现已向 Pro 和 Ultra 用户开放。东锡 NLP 引用该消息并指出,Neon 在金融、会计、电子表格建模和复杂约束遵循评测中多数 Benchmark 排名第一,其中 Legal Research Bench、Tax Agent Bench 和 AutomationBench 位列第一。

    引用Demis Hassabis@demishassabis

    Excited to launch Gemini 4 Neon: frontier-level intelligence, super fast task completion, and available to Pro & Ultra users now!! We think you’ll enjoy this model. Alchemy at its finest… 🧪

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Neon 模型》

  2. 🚨 AI News | TestingCatalog82

    Google 正式发布前沿模型 Gemini 4 Neon,并称其在多项基准上超过 Opus 5.5 和 GPT-6.2 Sol。该模型现已向 Google AI Pro 和 Ultra 订阅用户开放,覆盖 Gemini、Antigravity 和 API。作者引用的 Google 方介绍称,Neon 在智能、速度与可用性之间取得平衡,并已在 Google AI Studio 和 API 上线。

    引用Logan Kilpatrick@OfficialLoganK

    Introducing Gemini 4 Neon! Pushing the frontier across domains and punching above its model class in a meaningful way. To me, Neon feels like the right balance of intelligence, speed, and usability. Available now across @GoogleAIStudio, the API, and more!

    同一件事,精选收录的是《Google 发布 Gemini 4,上线 Gemini App、Antigravity 与 API》

11月18日周三
  1. Rohan Paul82

    Google 将 Gemini 的个人智能体功能改名为 Task,并在美国、日本、韩国、加拿大、澳大利亚和印度免费向数亿用户开放,该功能此前名为 Spark,需订阅每月 19.99 美元的 AI Pro 套餐。

    引用Sundar Pichai@sundarpichai

    Knowing things → doing things → finding ways to save you time and money We’re expanding Tasks in Gemini free of charge to hundreds of millions of people in six countries. Powered by Gemini 4 Hydrogen, Tasks lets you hand off everyday life admin directly: - Forward emails to gemini@google.com to RSVP and add it to your family calendar - Have Gemini call local businesses to compare price quotes and check inventory (with full transcripts + approval before dialing) - Check out securely via Google Pay (uses the Agent Payments Protocol - AP2) - Seamlessly sign in via Google Password Manager We wanted to make this helpful + secure, we’re also partnering directly with major websites and security providers so it works well for site owners and agents.

    同一件事,精选收录的是《Google 将 Spark 智能体更名为 Tasks 并免费开放》

11月13日周五
11月12日周四
  1. 宝玉82

    Google 在 Gemini API 中上线 Decisions API,配套新模型 Gemini Decision 1,不生成文字,直接返回每道题的答案和 0 到 1 的置信度。

    引用Sundar Pichai@sundarpichai

    Today we’re launching the Gemini Decisions API, powered by our new Gemini Decision 1 model in @GoogleAIStudio. Instead of generating text token by token, it’s built for closed-set choices scoring up to 1,000 questions across text, images, audio, video + PDFs in a single pass. A few details: - 3 decision types: yes/no, multiple choice (up to 255 options) + rubric scoring - $0.10 / 1M input tokens with $0 billed for output tokens (~5x cheaper than Flash-Lite) - leads on several benchmark evals with lowest calibration error Google teams are already building with it, below is a look at Stitch Rapid designing a full product flow in seconds (1x speed)

    同一件事,精选收录的是《Google 发布 Gemini Decision 1 与 Gemini Decisions API》

11月11日周三
  1. Andrew Curran82

    Google 的 Gemini 4 Carbon 今日上线,面向需要持续运行的长周期任务,覆盖复杂软件工程、网络安全和科学发现等场景。该模型通过 Gemini API 在 Google AI Studio 和 Gemini for business 全球开放,Ultra 订阅用户可在 Antigravity、Gemini App 和 Search 的 AI Mode 中使用。Andrew Curran 转发该消息并提到 Carbon 正向 Ultra 订阅用户推送。

    引用Sundar Pichai@sundarpichai

    Gemini 4 Carbon is rolling out today. It’s built for sustained, long-horizon tasks, reasoning deeply through complex software engineering, cybersecurity, and scientific discovery. Coming globally via the Gemini API in @GoogleAIStudio and Gemini for business, and for Ultra subscribers in @antigravity, the @GeminiApp, and AI Mode in Search.

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Carbon 模型》

10月29日周四
11月12日周四
  1. Andrew Curran80

    Google 发布 Gemini Decisions API,由新模型 Gemini Decision 1 驱动,面向封闭集合选择打分,单次可处理最多 1,000 道文本、图像、音频、视频和 PDF 题目。它支持是/否、最多 255 个选项的多选题和评分量表三种决策类型,输入定价 $0.10/1M tokens 且输出不计费,约为 Flash-Lite 的 5 倍便宜。Google 称其在多项基准评测中领先且校准误差最低,并已用于 Stitch Rapid 的产品流程设计。

    引用Sundar Pichai@sundarpichai

    Today we’re launching the Gemini Decisions API, powered by our new Gemini Decision 1 model in @GoogleAIStudio. Instead of generating text token by token, it’s built for closed-set choices scoring up to 1,000 questions across text, images, audio, video + PDFs in a single pass. A few details: - 3 decision types: yes/no, multiple choice (up to 255 options) + rubric scoring - $0.10 / 1M input tokens with $0 billed for output tokens (~5x cheaper than Flash-Lite) - leads on several benchmark evals with lowest calibration error Google teams are already building with it, below is a look at Stitch Rapid designing a full product flow in seconds (1x speed)

    同一件事,精选收录的是《Google 发布 Gemini Decision 1 与 Gemini Decisions API》

11月19日周四
11月12日周四
  1. 🚨 AI News | TestingCatalog78

    Google 推出 Gemini Decisions API,由新模型 Gemini Decision 1 驱动,已在 Google AI Studio 和 API 上线。该模型不做逐 token 文本生成,而是面向封闭集选择打分,单次可处理最多 1000 个问题,覆盖文本、图像、音频、视频和 PDF。支持是/否、最多 255 个选项的多选题和评分量表三种决策类型,定价为每 1M 输入 token $0.10、输出 token 不计费。引用内容称其在多项基准评测中领先且校准误差最低。

    引用Sundar Pichai@sundarpichai

    Today we’re launching the Gemini Decisions API, powered by our new Gemini Decision 1 model in @GoogleAIStudio. Instead of generating text token by token, it’s built for closed-set choices scoring up to 1,000 questions across text, images, audio, video + PDFs in a single pass. A few details: - 3 decision types: yes/no, multiple choice (up to 255 options) + rubric scoring - $0.10 / 1M input tokens with $0 billed for output tokens (~5x cheaper than Flash-Lite) - leads on several benchmark evals with lowest calibration error Google teams are already building with it, below is a look at Stitch Rapid designing a full product flow in seconds (1x speed)

    同一件事,精选收录的是《Google 发布 Gemini Decision 1 与 Gemini Decisions API》

10月29日周四
  1. 宝玉78

    Google 于 10 月 28 日发布新模型 Gemini 4 Carbon,距上一个模型 Gemini 4 Argon 仅隔两周,在 Artificial Analysis 综合排行榜 v5.0 上以 59 分排第一,比 Claude Opus 5.5 高 1 分、比 GPT-6 Astra 高 2 分。

    引用Sundar Pichai@sundarpichai

    I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model. Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards. As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!

    同一件事,精选收录的是《Google DeepMind 发布 Gemini 4 Carbon 前沿模型》

10月28日周三
  1. Riley Goodside78

    Google DeepMind 发布世界模型 Genie 4,称其为目前最真实、最具交互性的世界模型。用户可用自然语言实时调整天气、移动镜头或加入新元素,Gemini 在背后充当模拟的“大脑”,负责把请求转成动作、规划复杂事件并跟踪场景变化;Genie 4 还新增环境音效和可实时调整的人类对话。Riley Goodside 转发时称其相比 Genie 3 进步很大。

    引用Google AI@GoogleAI

    Introducing Genie 4 from @GoogleDeepMind, our most realistic and interactive world model yet 🌍 Now you can use natural language to dynamically steer the world and its soundscape in real-time, whether you're changing the weather, moving the camera, or introducing new elements to the environment. So, how does the model do that? Behind the scenes, Gemini acts as the "brain" of the simulation. It translates your requests into actions, plans out complex events, watches the scene unfold to keep track of what's happening, and adjusts as you change the story. Genie 4 also introduces ambient audio and real-time interactive human dialogue that adapts on the fly.

    同一件事,精选收录的是《Google DeepMind 发布世界模型 Genie 4》

10月21日周三
11月6日周五
11月19日周四