Tag
The paper introduces GUARD, a grader-free framework for stabilizing recursive self-improvement in LLM agents by decomposing tasks into pre-execution checks and guiding harness edits with directional regret, achieving continual gains without catastrophic forgetting or memory bloat while cutting token usage.
Google Cloud announces three data updates for AI agents — Knowledge Catalog, Borderless Lakehouse, and Smart Storage — aimed at simplifying data preparation without complex migrations or costly egress fees.
Alphabet的谷歌云第三季度收入翻倍至307亿美元,但搜索业务增长放缓且自由现金流转负,导致盘后股价暴跌逾11%,市值蒸发超2800亿美元。
微软首次披露Azure季度营收为332亿美元,同比增长48%,略高于Google Cloud的307亿美元,并超出分析师预期,云巨头之间的收入对比首次可以直接进行。
Google brings Gemini Robotics Embodied Reasoning 2 (ER 2), a specialized embodied reasoning model for robots, to public preview on Google Cloud's Gemini platform for enterprise deployments, enabling spatio-temporal reasoning, long-horizon planning, and multi-robot collaboration.
Google's TPUs will run on a hyperscale cloud outside Google for the first time, with Oracle deploying capacity from 2027. Google touts 8th-gen TPU performance gains and its open-source TorchTPU stack for PyTorch compatibility, expanding TPU sales into direct hyperscale competition.
Google Cloud AI Research, in collaboration with UIUC, proposes OO-Harness, which introduces object-oriented design (inheritance, encapsulation, polymorphism) into the automatic evolution of agent harnesses. Through a class hierarchy, it enables code reuse and safe fallback for specialized scaffolds. On three long-horizon benchmarks, the main metric improves from 61.62% to 77.41%, and it significantly outperforms traditional self-evolution methods on out-of-distribution tasks.
OpenAI and Google达成合作,OpenAI最新的GPT 6与6.1系列模型(Astra、Sol、Luna)将登陆Google Cloud的Model Garden,并集成进Google的编程工具Antigravity,使Google Cloud成为唯一同时提供三大闭源模型的云平台。
OpenAI and Google Cloud are partnering to bring OpenAI's frontier models and Codex agentic workflows into Google Cloud's Model Garden, available via Google Cloud Marketplace. General availability is planned in the coming weeks, while Microsoft remains OpenAI's primary cloud partner.
Google Cloud and NVIDIA announced the general availability of A5X instances powered by NVIDIA Vera Rubin NVL72, scaling to up to 960,000 Rubin GPUs for agentic inference and large-scale reinforcement learning, with early customers including Ineffable Intelligence and WPP.
Google released TorchTPU as an open source project, letting developers run native PyTorch apps on its TPUs without rewriting code, joining Amazon's TorchNeuron in challenging Nvidia's dominance over AI chip software tooling.
A Google Cloud outage disrupted banking services and numerous other operations across Europe, highlighting reliance on cloud infrastructure for critical financial systems.
The tweet reposted a tutorial billed as the first of its kind on the internet: using 60 lines of local code to bypass Pro restrictions in agy and use gemini-4-argon for free with Google Cloud’s $300 credit, while claiming that access could be renewed indefinitely by switching Google accounts.
Cloud Releases 是一款在 Product Hunt 上发布的产品,它将 AWS、Azure、Google Cloud 和 Cloudflare 的更新汇总到一个信息流中,让用户只关注自己使用的服务并及早发现退役公告。
Gemini 4 Argon 现已在 Manus 平台上线,由 Google Cloud 提供支持,支持代码编写、规划以及 3D 建模等多模态任务。
Google Cloud 在 Gemini at Work 2026 上发布 Gemini agent——一个面向工作场景的统一智能体,可在单一提示框中回答问题、处理知识工作、创建媒体、编写并运行代码,并深度集成 Google Workspace、行业专属技能与企业级安全治理能力。
Google Cloud 在 Gemini at Work 2026 活动上发布了 Gemini agent——一个统一的企业级工作智能体,可跨模型编排、连接企业数据与工作流,支持知识工作、图像生成及代码编写与运行。新功能还包括 Gemini 深度集成 Google Workspace、自然语言数据分析技能、行业定制方案、企业级安全治理与成本控制。
Simon Willison argues that usage-based services and APIs should offer default hard spending caps rather than soft warnings, as AI coding and personal agents make it easy to spin up costly runaway services. He notes AWS recently launched monthly spend limits and Google Cloud released Spend Caps, suggesting the trend is gaining momentum.
Season 3 of Google Cloud's Advent of Agents launches — a free 31-day online learning program teaching developers how to build and secure production-ready agentic systems, with daily tutorials and hands-on exercises.
Phil Schmid 报道 Gemini Enterprise Agent Platform(前身为 Vertex AI)现已支持 Interactions API 与 Gemini 模型集成,并继续沿用 `google-genai` SDK;该 API 提供统一的有状态接口,支持服务端会话状态、可观察的执行步骤与后台长任务执行。