https://www.gpts24.com/en/news/world-labs-debuts-frameboy-real-time-atlas-streaming-at-1440p-via-cloud-gpuGPTS24en2026-11-06T13:44:39.369ZWorld Labs Debuts Frameboy: Real-Time Atlas Streaming at 1440p via Cloud GPUhttps://www.gpts24.com/zh/news/agentic-robots-survey-why-getting-motion-right-is-no-longer-enough-to-ship-a-robotGPTS24zh-cn2026-11-06T01:33:23.195Z智能体机器人综述：光是动作对了，已经不足以让一台机器人真正能用https://www.gpts24.com/en/news/agentic-robots-survey-why-getting-motion-right-is-no-longer-enough-to-ship-a-robotGPTS24en2026-11-06T01:33:23.195ZAgentic Robots Survey: Why Getting Motion Right Is No Longer Enough to Ship a Robothttps://www.gpts24.com/zh/news/claude-code-product-lead-cat-wu-leaves-anthropic-for-openai-core-productsGPTS24zh-cn2026-11-06T01:32:41.673ZClaude Code 产品负责人 Cat Wu 离开 Anthropic，加入 OpenAI 核心产品部门https://www.gpts24.com/en/news/claude-code-product-lead-cat-wu-leaves-anthropic-for-openai-core-productsGPTS24en2026-11-06T01:32:41.673ZClaude Code Product Lead Cat Wu Leaves Anthropic for OpenAI Core Productshttps://www.gpts24.com/zh/news/aformic-lands-6-95m-to-scale-amr-fleet-serving-toyota-3m-and-defense-oemsGPTS24zh-cn2026-11-05T21:48:23.122ZAformic 融资 695 万美元，扩大服务丰田、3M 和国防承包商的自主移动机器人机群https://www.gpts24.com/en/news/aformic-lands-6-95m-to-scale-amr-fleet-serving-toyota-3m-and-defense-oemsGPTS24en2026-11-05T21:48:23.122ZAformic Lands $6.95M to Scale AMR Fleet Serving Toyota, 3M, and Defense OEMshttps://www.gpts24.com/zh/news/openai-commits-to-full-power-costs-and-bans-official-ndas-followingGPTS24zh-cn2026-11-05T21:47:44.857Z中期选举反弹之后，OpenAI 承诺承担全部电力成本，并禁止官方保密协议https://www.gpts24.com/en/news/openai-commits-to-full-power-costs-and-bans-official-ndas-followingGPTS24en2026-11-05T21:47:44.857ZOpenAI Commits to Full Power Costs and Bans Official NDAs Following Data-Center Election Backlashhttps://www.gpts24.com/zh/news/ntt-data-commits-900-mw-to-bloom-energy-fuel-cells-for-ai-data-center-primary-powerGPTS24zh-cn2026-11-05T21:46:57.801ZNTT DATA 承诺采购 900 兆瓦 Bloom Energy 燃料电池，用作 AI 数据中心的主电源https://www.gpts24.com/en/news/ntt-data-commits-900-mw-to-bloom-energy-fuel-cells-for-ai-data-center-primary-powerGPTS24en2026-11-05T21:46:57.801ZNTT DATA Commits 900 MW to Bloom Energy Fuel Cells for AI Data Center Primary Powerhttps://www.gpts24.com/zh/news/ibm-supermicro-and-nvidia-validate-kv-cache-storage-that-cuts-inference-latency-by-56xGPTS24zh-cn2026-11-05T20:53:51.151ZIBM、超微和英伟达共同验证一套 KV 缓存存储方案，把推理延迟砍到 1/56https://www.gpts24.com/en/news/ibm-supermicro-and-nvidia-validate-kv-cache-storage-that-cuts-inference-latency-by-56xGPTS24en2026-11-05T20:53:51.151ZIBM, Supermicro, and NVIDIA Validate KV Cache Storage That Cuts Inference Latency by 56xhttps://www.gpts24.com/zh/news/spacex-targets-13gw-of-ai-data-centers-by-2027-backed-by-40b-gpu-financing-dealGPTS24zh-cn2026-11-05T20:52:50.918ZSpaceX 目标到 2027 年建成 13 吉瓦 AI 数据中心，背后是一笔 400 亿美元的 GPU 融资交易https://www.gpts24.com/en/news/spacex-targets-13gw-of-ai-data-centers-by-2027-backed-by-40b-gpu-financing-dealGPTS24en2026-11-05T20:52:50.918ZSpaceX Targets 13GW of AI Data Centers by 2027, Backed by $40B GPU Financing Dealhttps://www.gpts24.com/zh/news/openai-pipeline-finds-948-linux-root-exploits-for-1-2m-aiGPTS24zh-cn2026-11-05T20:39:29.268ZOpenAI 的流水线花 120 万美元，找出 948 个 Linux 提权漏洞：AI 把内核安全研究变成了流水线作业https://www.gpts24.com/en/news/openai-pipeline-finds-948-linux-root-exploits-for-1-2m-aiGPTS24en2026-11-05T20:39:29.268ZOpenAI Pipeline Finds 948 Linux Root Exploits for $1.2M: AI Industrializes Kernel Security Researchhttps://www.gpts24.com/zh/news/disney-tests-gps-free-robots-in-parking-lots-as-imagineering-eyes-character-applicationsGPTS24zh-cn2026-11-05T20:35:57.328Z迪士尼在停车场测试无 GPS 机器人，Imagineering 同时考察角色机器人应用https://www.gpts24.com/en/news/disney-tests-gps-free-robots-in-parking-lots-as-imagineering-eyes-character-applicationsGPTS24en2026-11-05T20:35:57.328ZDisney Tests GPS-Free Robots in Parking Lots as Imagineering Eyes Character Applicationshttps://www.gpts24.com/zh/news/cognition-swe-2-flash-runs-3-faster-at-one-fifth-the-cost-of-its-flagship-coding-modelGPTS24zh-cn2026-11-05T19:42:07.325ZCognition 的 SWE-2 Flash 速度快 3 倍，成本只是旗舰编程模型的五分之一https://www.gpts24.com/en/news/cognition-swe-2-flash-runs-3-faster-at-one-fifth-the-cost-of-its-flagship-coding-modelGPTS24en2026-11-05T19:42:07.325ZCognition SWE-2 Flash Runs 3× Faster at One-Fifth the Cost of Its Flagship Coding Modelhttps://www.gpts24.com/zh/news/chatgpt-images-2-6-delivers-gif-generation-as-openai-retreats-from-video-aiGPTS24zh-cn2026-11-05T19:40:21.212ZChatGPT Images 2.6 带来 GIF 生成能力，OpenAI 同时从视频 AI 上后退https://www.gpts24.com/en/news/chatgpt-images-2-6-delivers-gif-generation-as-openai-retreats-from-video-aiGPTS24en2026-11-05T19:40:21.212ZChatGPT Images 2.6 Delivers GIF Generation as OpenAI Retreats From Video AIhttps://www.gpts24.com/zh/news/baidu-releases-280b-ernie-5-5-flash-base-restricted-license-excludes-eu-and-ukGPTS24zh-cn2026-11-05T19:39:17.228Z百度发布 2800 亿参数的 ERNIE 5.5 Flash Base：限制性许可排除欧盟和英国https://www.gpts24.com/en/news/baidu-releases-280b-ernie-5-5-flash-base-restricted-license-excludes-eu-and-ukGPTS24en2026-11-05T19:39:17.228ZBaidu Releases 280B ERNIE 5.5 Flash Base: Restricted License Excludes EU and UKhttps://www.gpts24.com/zh/news/nvidia-research-ai-agents-hold-accuracy-at-16k-tokens-as-long-contexts-cost-7x-moreGPTS24zh-cn2026-11-05T19:37:53.024Z英伟达研究：AI 智能体在 1.6 万词元上仍保持准确率，长上下文反而贵 7 倍https://www.gpts24.com/en/news/nvidia-research-ai-agents-hold-accuracy-at-16k-tokens-as-long-contexts-cost-7x-moreGPTS24en2026-11-05T19:37:53.024ZNVIDIA Research: AI Agents Hold Accuracy at 16K Tokens as Long Contexts Cost 7x Morehttps://www.gpts24.com/zh/news/ricursive-intelligence-claims-us-coding-ai-lead-with-dense-65b-modelGPTS24zh-cn2026-11-05T19:35:46.503ZRicursive Intelligence 宣称以一个 650 亿参数的稠密模型，拿下美国编程 AI 领先地位，成本只是对手的一个零头https://www.gpts24.com/en/news/ricursive-intelligence-claims-us-coding-ai-lead-with-dense-65b-modelGPTS24en2026-11-05T19:35:46.503ZRicursive Intelligence Claims US Coding AI Lead With Dense 65B Model at a Fraction of Rivals' Costhttps://www.gpts24.com/zh/news/openai-hires-firm-that-breached-anthropic-s-systems-to-lead-internal-security-sprintGPTS24zh-cn2026-11-05T14:04:10.544ZOpenAI 聘用了那家攻破 Anthropic 系统的公司，来主导内部安全冲刺https://www.gpts24.com/en/news/openai-hires-firm-that-breached-anthropic-s-systems-to-lead-internal-security-sprintGPTS24en2026-11-05T14:04:10.544ZOpenAI Hires Firm That Breached Anthropic's Systems to Lead Internal Security Sprinthttps://www.gpts24.com/zh/news/anthropic-open-sources-flashprove-to-cryptographically-verify-ai-model-identityGPTS24zh-cn2026-11-05T13:42:41.125ZAnthropic 开源 FlashProve，用密码学方式验证 AI 模型身份https://www.gpts24.com/en/news/anthropic-open-sources-flashprove-to-cryptographically-verify-ai-model-identityGPTS24en2026-11-05T13:42:41.125ZAnthropic Open-Sources FlashProve to Cryptographically Verify AI Model Identityhttps://www.gpts24.com/zh/news/openai-locks-demand-response-into-every-future-ai-campus-with-grid-protectors-programGPTS24zh-cn2026-11-05T13:42:08.201ZOpenAI 推出 Grid Protectors 计划，把需求响应写进未来每一座 AI 园区https://www.gpts24.com/en/news/openai-locks-demand-response-into-every-future-ai-campus-with-grid-protectors-programGPTS24en2026-11-05T13:42:08.201ZOpenAI Locks Demand Response Into Every Future AI Campus With Grid Protectors Programhttps://www.gpts24.com/zh/news/amd-posts-record-13-4b-quarter-as-instinct-mi450-pushes-data-center-to-8-5bGPTS24zh-cn2026-11-05T13:41:30.947ZAMD 单季营收创纪录达 134 亿美元，Instinct MI450 把数据中心业务推上 85 亿美元https://www.gpts24.com/en/news/amd-posts-record-13-4b-quarter-as-instinct-mi450-pushes-data-center-to-8-5bGPTS24en2026-11-05T13:41:30.947ZAMD Posts Record $13.4B Quarter as Instinct MI450 Pushes Data Center to $8.5Bhttps://www.gpts24.com/zh/news/big-tech-s-off-balance-sheet-ai-debt-hits-3-2-trillion-as-shadow-obligations-doubleGPTS24zh-cn2026-11-05T13:11:06.247Z大科技公司表外 AI 债务冲到 3.2 万亿美元，影子负债半年翻倍https://www.gpts24.com/en/news/big-tech-s-off-balance-sheet-ai-debt-hits-3-2-trillion-as-shadow-obligations-doubleGPTS24en2026-11-05T13:11:06.247ZBig Tech's Off-Balance-Sheet AI Debt Hits $3.2 Trillion as Shadow Obligations Double