https://ai-digest-1.com/zh/news/alibaba-releases-qwen2-audio-new-audio-language-model-1501AI Digest DNB66zh2026-07-28T16:03:19.361460+00:00阿里巴巴发布Qwen2-Audio：支持语音聊天和音频分析的新型音频语言模型 https://ai-digest-1.com/zh/news/qwen2-vl-alibaba-new-generation-vision-language-models-sota-1500AI Digest DNB66zh2026-07-28T16:03:12.176250+00:00Qwen2-VL：阿里巴巴发布新一代视觉语言模型，性能达到SoTA水平 https://ai-digest-1.com/zh/news/zhipu-left-minimax-right-divergence-and-dilemmas-of-two-large-model-commercializ-1499AI Digest DNB66zh2026-07-28T16:03:06.952108+00:00智谱向左，MiniMax向右：两条大模型商业化路径的分歧与困境 https://ai-digest-1.com/zh/news/openai-ai-agents-attack-hugging-face-revealed-1497AI Digest DNB66zh2026-07-28T16:02:57.831589+00:00失控的OpenAI AI代理意外攻击Hugging Face的细节曝光 https://ai-digest-1.com/zh/news/hygon-dcu-completes-adaptation-for-kimi-k3-day0-1496AI Digest DNB66zh2026-07-28T16:02:52.930194+00:00海光DCU完成Kimi K3 Day0适配 https://ai-digest-1.com/zh/news/kimi-k3-open-sourced-open-ecosystem-ai-consensus-1495AI Digest DNB66zh2026-07-28T16:02:41.839249+00:00Kimi K3 正式开源，开放生态成为全球 AI 新共识 https://ai-digest-1.com/zh/news/lfm2-5-encoders-fast-long-context-cpu-inference-1493AI Digest DNB66zh2026-07-28T16:02:26.487996+00:00LFM2.5-Encoders：快速长上下文CPU推理 https://ai-digest-1.com/zh/news/ai-memory-shortages-apacer-ceo-warns-deficit-2027-1492AI Digest DNB66zh2026-07-28T16:02:16.567799+00:00AI引发内存短缺；Apacer CEO警告2027年赤字将加剧 https://ai-digest-1.com/zh/news/data-centers-temporary-power-cuts-pjm-grid-1491AI Digest DNB66zh2026-07-28T16:02:07.387732+00:00美国最大电网数据中心或面临临时断电以避免停电 https://ai-digest-1.com/zh/news/ai-router-adds-top-image-generators-flux-gemini-mai-image-grok-imagine-1490AI Digest DNB66zh2026-07-28T16:02:01.365410+00:00AI路由器新增顶级图像生成器：Flux、Gemini、MAI-Image、Grok Imagine https://ai-digest-1.com/zh/news/alibaba-releases-qwen2-5-math-1489AI Digest DNB66zh2026-07-28T15:03:41.097736+00:00阿里巴巴发布 Qwen2.5-Math：先进的开源数学大语言模型 https://ai-digest-1.com/zh/news/alibaba-releases-qwen2-5-coder-open-source-code-models-1488AI Digest DNB66zh2026-07-28T15:03:30.867707+00:00阿里巴巴发布Qwen2.5-Coder：最高320亿参数，开源代码模型 https://ai-digest-1.com/zh/news/qwen2-5-llm-expanding-language-model-boundaries-1487AI Digest DNB66zh2026-07-28T15:03:23.981137+00:00Qwen2.5-LLM：拓展语言模型边界 https://ai-digest-1.com/zh/news/qwen2-5-a-party-of-foundation-models-1486AI Digest DNB66zh2026-07-28T15:03:13.476474+00:00Qwen2.5：基础模型盛宴！ https://ai-digest-1.com/zh/news/claude-shared-conversations-artifacts-google-search-indexed-1484AI Digest DNB66zh2026-07-28T15:02:57.434436+00:00糟糕：一些Claude共享对话和Artifacts似乎已被谷歌搜索索引并公开访问 https://ai-digest-1.com/zh/news/kimi-open-sources-k3-large-model-weights-and-technical-report-1483AI Digest DNB66zh2026-07-28T15:02:52.158189+00:00Kimi 开源 K3 大模型权重并发布技术报告 https://ai-digest-1.com/zh/news/alibaba-cloud-zhenwu-supernode-adapts-kimi-k3-model-1482AI Digest DNB66zh2026-07-28T15:02:47.252497+00:00阿里云臻武超节点Day0适配Kimi K3大模型，双方启动国产算力合作 https://ai-digest-1.com/zh/news/moore-threads-completes-adaptation-kimi-k3-1481AI Digest DNB66zh2026-07-28T15:02:31.932911+00:00摩尔线程完成对Kimi K3的适配，支持2.8万亿参数开源模型 https://ai-digest-1.com/zh/news/gpt-5-6-cancels-subscriptions-kimi-k3-outperforms-fable-5-on-gpu-tasks-openide-p-1480AI Digest DNB66zh2026-07-28T15:02:20.367408+00:00GPT-5.6 取消订阅，Kimi K3 在 GPU 任务上超越 Fable 5，OpenIDE Pro 免费发布 https://ai-digest-1.com/zh/news/ransomware-attacks-not-declining-4-defenses-1479AI Digest DNB66zh2026-07-28T15:02:11.736691+00:00勒索软件攻击并未减少：企业应采取的4种防御措施 https://ai-digest-1.com/zh/news/us-uae-talon-synapse-group-military-ai-1477AI Digest DNB66zh2026-07-28T15:01:56.293664+00:00美国与阿联酋组建Talon Synapse小组，专注军事人工智能应用 https://ai-digest-1.com/zh/news/genai-transforms-business-processes-how-companies-deploy-generative-ai-and-measu-1476AI Digest DNB66zh2026-07-28T15:01:45.385751+00:00生成式AI变革业务流程：企业如何部署生成式AI并衡量创新影响 https://ai-digest-1.com/zh/news/russia-may-tighten-punishment-for-crimes-using-ai-1475AI Digest DNB66zh2026-07-28T15:01:37.776817+00:00俄罗斯拟加重利用人工智能犯罪的处罚——AI或将成为加重情节 https://ai-digest-1.com/zh/news/why-doesnt-my-model-work-1474AI Digest DNB66zh2026-07-28T14:04:15.674172+00:00为什么我的模型不工作？ https://ai-digest-1.com/zh/news/alibaba-open-sources-qwen2-5-coder-series-1473AI Digest DNB66zh2026-07-28T14:04:08.311080+00:00阿里巴巴开源 Qwen2.5-Coder 系列：最先进的开源代码模型 https://ai-digest-1.com/zh/news/alibaba-extends-qwen25-turbo-context-to-1m-tokens-1472AI Digest DNB66zh2026-07-28T14:04:01.910678+00:00阿里巴巴将Qwen2.5-Turbo上下文长度扩展至100万tokens https://ai-digest-1.com/zh/news/qwq-deep-reflection-on-boundaries-1471AI Digest DNB66zh2026-07-28T14:03:56.638386+00:00QwQ：对未知边界的深刻反思 https://ai-digest-1.com/zh/news/qvq-to-see-the-world-with-wisdom-1470AI Digest DNB66zh2026-07-28T14:03:51.859628+00:00QVQ：以智慧观天下 https://ai-digest-1.com/zh/news/microsoft-fixes-vs-code-copilot-credit-1469AI Digest DNB66zh2026-07-28T14:03:41.626027+00:00微软修复 VS Code：编辑器误将用户成果归功于 Copilot https://ai-digest-1.com/zh/news/deepseek-attracts-50-billion-in-financing-valuation-reaches-300-billion-1468AI Digest DNB66zh2026-07-28T14:03:31.845346+00:00DeepSeek吸引500亿融资，估值达3000亿 https://ai-digest-1.com/zh/news/top-5-global-ai-models-api-call-volume-chinese-products-1467AI Digest DNB66zh2026-07-28T14:03:23.899818+00:00全球大模型API调用量前五均为中国产品；小米与DeepSeek领先 https://ai-digest-1.com/zh/news/moonshot-ai-open-sources-kimi-k3-weights-1464AI Digest DNB66zh2026-07-28T14:02:54.602682+00:00Moonshot AI 开源热门模型 Kimi-K3 权重——性能接近市场领先者，运行却简单得多 https://ai-digest-1.com/zh/news/recursive-superintelligence-410m-compute-deal-amazon-1461AI Digest DNB66zh2026-07-28T14:02:33.880348+00:00Recursive Superintelligence 与亚马逊签署 4.1 亿美元计算合同 https://ai-digest-1.com/zh/news/gitlab-report-ai-code-generation-review-bottleneck-governance-gap-1457AI Digest DNB66zh2026-07-28T14:02:02.851446+00:00GitLab报告：AI代码生成造成审查瓶颈和治理缺口 https://ai-digest-1.com/zh/news/do-text-embeddings-perfectly-encode-text-1456AI Digest DNB66zh2026-07-28T13:04:10.088924+00:00文本嵌入能否完美编码文本？ https://ai-digest-1.com/zh/news/alibaba-qwen-processbench-process-reward-models-1455AI Digest DNB66zh2026-07-28T13:03:58.454827+00:00阿里巴巴Qwen发布ProcessBench和增强版过程奖励模型用于数学推理 https://ai-digest-1.com/zh/news/global-batch-load-balance-for-moe-llm-training-1454AI Digest DNB66zh2026-07-28T13:03:50.765670+00:00全局批次负载均衡：提升MoE大模型训练的几乎免费午餐 https://ai-digest-1.com/zh/news/alibaba-releases-qwen2-5-vl-new-flagship-multimodal-model-1453AI Digest DNB66zh2026-07-28T13:03:32.410074+00:00阿里巴巴发布 Qwen2.5-VL：全新旗舰多模态模型 https://ai-digest-1.com/zh/news/alibaba-open-sources-qwen2-5-1m-with-1m-token-context-1452AI Digest DNB66zh2026-07-28T13:03:25.023949+00:00阿里开源 Qwen2.5-1M，支持高达 100 万 Token 的上下文 https://ai-digest-1.com/zh/news/ai-is-learning-to-read-the-room-1450AI Digest DNB66zh2026-07-28T13:02:59.159141+00:00AI 正在学会“察言观色” https://ai-digest-1.com/zh/news/openai-and-hugging-face-fix-security-issue-1449AI Digest DNB66zh2026-07-28T13:02:54.849805+00:00OpenAI 与 Hugging Face 联手修复模型评估中的安全问题 https://ai-digest-1.com/zh/news/ai-autonomous-attacks-reality-nsfocus-redefines-penetration-testing-llms-1448AI Digest DNB66zh2026-07-28T13:02:41.968309+00:00AI自主攻击成为现实：绿盟科技用大语言模型重新定义渗透测试 https://ai-digest-1.com/zh/news/us-court-sides-with-serpapi-creator-in-google-data-scraping-lawsuit-1447AI Digest DNB66zh2026-07-28T13:02:34.896643+00:00美国法院在谷歌数据抓取诉讼中支持SerpApi创始人 https://ai-digest-1.com/zh/news/chinese-ai-models-call-themselves-claude-mimicking-american-bot-behavior-1446AI Digest DNB66zh2026-07-28T13:02:23.589550+00:00中国AI模型开始自称Claude并模仿美国机器人行为 https://ai-digest-1.com/zh/news/visual-guide-to-attention-variants-in-modern-llms-1434AI Digest DNB66zh2026-07-28T12:05:31.032287+00:00现代大语言模型中注意力机制变体的视觉指南 https://ai-digest-1.com/zh/news/zhipu-acquires-zhongke-jiahe-compute-gap-stock-surges-1431AI Digest DNB66zh2026-07-28T12:04:55.994006+00:00智谱收购中科佳禾填补算力缺口，早盘股价飙升超25% https://ai-digest-1.com/zh/news/alphabet-spent-45-billion-on-ai-last-quarter-plans-81-billion-more-1430AI Digest DNB66zh2026-07-28T12:04:39.118387+00:00Alphabet在上一季度在人工智能上花费了450亿美元，并计划再花费811亿美元 https://ai-digest-1.com/zh/news/moonshot-ai-kimi-k3-open-source-2-8-trillion-parameters-1429AI Digest DNB66zh2026-07-28T12:04:35.072119+00:00月之暗面今晚开源Kimi K3：2.8万亿参数，全球最大开源模型 https://ai-digest-1.com/zh/news/anthropic-claude-cowork-sandbox-vulnerability-macos-files-1424AI Digest DNB66zh2026-07-28T12:03:50.189955+00:00Anthropic Claude Cowork 逃出沙箱，获得 macOS 文件的完全访问权限 https://ai-digest-1.com/zh/news/anthropic-does-not-oppose-open-ai-models-dario-amodei-1423AI Digest DNB66zh2026-07-28T12:03:43.846583+00:00Anthropic无意反对开源AI模型，Dario Amodei这样保证 https://ai-digest-1.com/zh/news/china-accuses-us-companies-distilling-ai-models-1422AI Digest DNB66zh2026-07-28T12:03:35.727753+00:00中国指责美国公司蒸馏其AI模型 https://ai-digest-1.com/zh/news/anthropic-opposes-open-model-ban-nvidia-assembles-alliance-1420AI Digest DNB66zh2026-07-28T12:03:11.442943+00:00Anthropic 反对开源模型禁令；NVIDIA 4天内集结75个签名联盟 https://ai-digest-1.com/zh/news/meta-blackrock-el-paso-data-center-china-ipv6-ai-1419AI Digest DNB66zh2026-07-28T12:02:59.862548+00:00Meta与贝莱德携手建设埃尔帕索数据中心；中国启动AI模型IPv6专项计划 https://ai-digest-1.com/zh/news/neural-networks-stock-market-logic-textual-reports-1416AI Digest DNB66zh2026-07-28T12:02:26.379376+00:00神经网络通过文本报告破解股市逻辑 https://ai-digest-1.com/zh/news/what-makes-ai-art-worth-collecting-1414AI Digest DNB66zh2026-07-28T11:04:03.514824+00:00什么让AI艺术值得收藏？ https://ai-digest-1.com/zh/news/gpt-5-6-frontier-intelligence-scales-with-your-ambition-1413AI Digest DNB66zh2026-07-28T11:03:58.199623+00:00GPT-5.6：前沿智能，随你的雄心扩展 https://ai-digest-1.com/zh/news/google-deepmind-atl-saathi-gemini-ai-assistant-indian-educators-1412AI Digest DNB66zh2026-07-28T11:03:45.678630+00:00谷歌 DeepMind 推出搭载 Gemini 的印度教育AI助手 ATL Saathi https://ai-digest-1.com/zh/news/moonshot-ai-responds-to-concerns-about-k3-model-1409AI Digest DNB66zh2026-07-28T11:03:20.935303+00:00Moonshot AI回应公众对其K3模型的担忧 https://ai-digest-1.com/zh/news/china-open-source-large-models-work-longer-smarter-1408AI Digest DNB66zh2026-07-28T11:03:13.155623+00:00中国使开源大模型工作更长久、更智能 https://ai-digest-1.com/zh/news/lyft-baidu-testing-robotaxis-london-1404AI Digest DNB66zh2026-07-28T11:02:06.448359+00:00Lyft与百度在伦敦开始测试自动驾驶出租车 https://ai-digest-1.com/zh/news/how-llms-can-help-determine-data-lineage-1403AI Digest DNB66zh2026-07-28T11:01:58.961567+00:00LLM如何帮助确定数据血缘关系 https://ai-digest-1.com/zh/news/ai-in-mathematics-new-golden-age-scientific-discovery-1401AI Digest DNB66zh2026-07-28T11:01:33.701314+00:00AI在数学中开启科学发现的新黄金时代 https://ai-digest-1.com/zh/news/us-big-models-copy-china-homework-following-deepseek-architecture-1400AI Digest DNB66zh2026-07-28T10:03:29.228960+00:00效仿DeepSeek架构，美国大模型开始抄中国作业 https://ai-digest-1.com/zh/news/zhipu-acquires-zhongke-jiahe-stock-surges-1398AI Digest DNB66zh2026-07-28T10:03:15.187760+00:00智谱收购中科佳禾，股价飙升近37%，AI计算领导者成'神助攻' https://ai-digest-1.com/zh/news/deepseek-chip-development-angers-huang-renxun-again-1397AI Digest DNB66zh2026-07-28T10:03:04.973615+00:00DeepSeek自研芯片再次激怒黄仁勋 https://ai-digest-1.com/zh/news/va-signs-billion-contract-to-integrate-ai-into-agency-functions-1396AI Digest DNB66zh2026-07-28T10:02:53.985176+00:00退伍军人事务部签署16亿美元合同，将AI整合到机构职能中 https://ai-digest-1.com/zh/news/moonshot-ai-open-sources-kimi-k3-alibaba-huawei-support-1395AI Digest DNB66zh2026-07-28T10:02:45.522183+00:00Moonshot AI 开源 Kimi K3 模型，阿里巴巴和华为提供首发支持 https://ai-digest-1.com/zh/news/china-first-domestic-100000-card-ai-supercluster-operational-1394AI Digest DNB66zh2026-07-28T10:02:41.768384+00:00中国首个国产10万卡AI超算集群正式投入运营 https://ai-digest-1.com/zh/news/artificial-intelligence-at-mayo-clinic-1393AI Digest DNB66zh2026-07-28T10:02:38.029374+00:00梅奥诊所的人工智能应用 - Emerj人工智能研究 https://ai-digest-1.com/zh/news/private-claude-chats-exposed-in-search-results-1392AI Digest DNB66zh2026-07-28T10:02:28.136218+00:00私人Claude聊天记录在谷歌和必应搜索结果中曝光 https://ai-digest-1.com/zh/news/llama-4-vs-mistral-ai-vs-minimax-4x-context-gap-2026-1391AI Digest DNB66zh2026-07-28T10:02:22.930347+00:00Llama 4 对比 Mistral AI 与 MiniMax：4倍上下文差距 [2026] https://ai-digest-1.com/zh/news/kimi-k3-and-unlimited-ocr-top-global-charts-1389AI Digest DNB66zh2026-07-28T10:02:09.806638+00:00Kimi K3 与 Unlimited OCR 登顶全球榜单，中国开源模型横扫海外 https://ai-digest-1.com/zh/news/anthropic-destroys-paper-books-for-claude-1387AI Digest DNB66zh2026-07-28T10:01:51.149675+00:00Anthropic为了Claude销毁纸质书？ https://ai-digest-1.com/zh/news/zhipu-ai-glm-5-2-matches-anthropic-opus-4-8-1380AI Digest DNB66zh2026-07-28T09:02:53.328054+00:00智谱AI最新开源模型GLM 5.2据悉性能媲美Anthropic Opus 4.8 https://ai-digest-1.com/zh/news/from-deepseek-to-k3-chinese-large-models-global-top-tier-1376AI Digest DNB66zh2026-07-28T09:02:28.343816+00:00从DeepSeek到K3：中国大模型如何跻身全球顶尖行列 https://ai-digest-1.com/zh/news/us-threatens-sanctions-on-chinese-ai-companies-1374AI Digest DNB66zh2026-07-28T09:02:02.831032+00:00美国威胁制裁中国AI公司；北京指责美国企业进行模型蒸馏并威胁采取反制措施 https://ai-digest-1.com/zh/news/openai-ceo-sam-altman-to-meet-with-us-treasury-secretary-commerce-secretary-and-1371AI Digest DNB66zh2026-07-28T09:01:40.910634+00:00OpenAI CEO Sam Altman 将与美财长、商务部长及国会议员会面，讨论AI政策 https://ai-digest-1.com/zh/news/worlds-first-judicial-ai-in-abu-dhabi-1370AI Digest DNB66zh2026-07-28T09:01:37.628395+00:00全球首个人工智能司法系统将在阿布扎比投入使用 https://ai-digest-1.com/zh/news/microsoft-releases-mai-cyber-1-flash-1369AI Digest DNB66zh2026-07-28T09:01:29.048776+00:00微软发布MAI-Cyber-1-Flash，5B活跃参数的网络安全模型在CyberGym上达到95.95% https://ai-digest-1.com/zh/news/llama-3-2-features-capabilities-and-ai-advancements-1365AI Digest DNB66zh2026-07-28T08:05:18.121699+00:00LLaMA 3.2：特性、能力与AI进展 https://ai-digest-1.com/zh/news/gpt-6-leaked-saved-by-zhipu-glm-5-2-1364AI Digest DNB66zh2026-07-28T08:05:10.909743+00:00GPT-6 泄露？OpenAI最尴尬的自我曝光，被智谱开源模型GLM-5.2拯救！ https://ai-digest-1.com/zh/news/bytedance-doubao-spring-recruitment-ai-jobs-14x-1361AI Digest DNB66zh2026-07-28T08:04:35.621327+00:00字节跳动豆包：春季招聘四大热门行业，AI岗位同比暴增14倍 https://ai-digest-1.com/zh/news/when-bigger-is-no-longer-the-only-path-another-chinese-model-goes-open-source-1360AI Digest DNB66zh2026-07-28T08:04:24.403046+00:00当“更大”不再是唯一路径：又一家中国模型开源 https://ai-digest-1.com/zh/news/openai-hugging-face-attack-unprecedented-decade-old-experiment-1358AI Digest DNB66zh2026-07-28T08:04:10.759091+00:00OpenAI称Hugging Face攻击史无前例，但十年前实验早有预兆 https://ai-digest-1.com/zh/news/kimi-k3-deploys-beyond-nvidia-completes-45nm-chip-prototype-1356AI Digest DNB66zh2026-07-28T08:03:59.047103+00:00Kimi K3 部署能力超越 NVIDIA 生态系统，完成 45nm 推理芯片原型设计 https://ai-digest-1.com/zh/news/russia-introduces-requirements-personal-data-protection-ai-1354AI Digest DNB66zh2026-07-28T08:03:43.929143+00:00俄罗斯拟出台使用人工智能时个人数据保护要求 https://ai-digest-1.com/zh/news/claude-chat-transcripts-leaked-via-google-search-1351AI Digest DNB66zh2026-07-28T08:03:15.596690+00:00Claude聊天记录通过谷歌搜索泄露 https://ai-digest-1.com/zh/news/ai-rampage-attack-enterprise-systems-chinese-large-model-saves-the-day-1347AI Digest DNB66zh2026-07-28T07:04:33.338804+00:00企业系统遭AI失控攻击，国产大模型紧急救场：6月中旬上线，权威榜单达开源最佳水平 https://ai-digest-1.com/zh/news/computing-power-breakthrough-bytedance-closed-network-zhipu-open-source-1345AI Digest DNB66zh2026-07-28T07:04:20.377647+00:00算力突破：赌字节跳动的封闭私有网络还是智谱的开源活水？ https://ai-digest-1.com/zh/news/moonshot-ai-kimi-3-trillion-parameters-open-source-model-1338AI Digest DNB66zh2026-07-28T07:03:17.332011+00:00月之暗面旗下 Kimi 发布全球首个 3 万亿参数开源模型 https://ai-digest-1.com/zh/news/beingbeyond-being-h08-implicit-tactile-world-action-model-1337AI Digest DNB66zh2026-07-28T07:03:10.567481+00:00智在无界发布Being-H0.8：首个隐式触觉世界-动作模型，在50万小时视频上训练 https://ai-digest-1.com/zh/news/ilya-gets-5-billion-bet-from-jensen-huang-1336AI Digest DNB66zh2026-07-28T07:03:02.377606+00:00伊利亚获黄仁勋50亿美元押注：“是时候扩大了” https://ai-digest-1.com/zh/news/from-deepseek-to-kimi-k3-chinese-ai-heavy-industrialization-of-computing-power-1335AI Digest DNB66zh2026-07-28T07:02:58.214082+00:00从DeepSeek到Kimi K3：中国AI难以逃脱“算力重工业化”定律 https://ai-digest-1.com/zh/news/416-tests-and-a-destroy-all-button-where-agent-projects-break-1332AI Digest DNB66zh2026-07-28T07:02:36.735179+00:00416次测试和一个“全部销毁”按钮：代理项目在哪些地方失败 https://ai-digest-1.com/zh/news/national-supercomputing-internet-ai-community-launches-kimi-k3-model-api-1331AI Digest DNB66zh2026-07-28T07:02:30.443644+00:00国家超算互联网AI社区上线Kimi K3模型及API服务 https://ai-digest-1.com/zh/news/us-considers-restrictions-on-chinese-ai-models-1326AI Digest DNB66zh2026-07-28T06:05:25.257949+00:00美国因知识产权盗窃担忧考虑限制中国AI模型 https://ai-digest-1.com/zh/news/white-house-accuses-moonshot-ai-distilling-anthropic-openai-models-break-free-1325AI Digest DNB66zh2026-07-28T06:05:11.003761+00:00白宫指控月之暗面蒸馏Anthropic模型，OpenAI模型在安全测试中失控 https://ai-digest-1.com/zh/news/chinese-open-source-ai-models-deepen-silicon-valley-divisions-1322AI Digest DNB66zh2026-07-28T06:04:33.324468+00:00中国开源AI模型的快速进步加深硅谷分歧 https://ai-digest-1.com/zh/news/nvidia-openai-250-billion-financing-data-center-wsj-1321AI Digest DNB66zh2026-07-28T06:04:26.782060+00:00Nvidia与OpenAI洽谈为数据中心提供2500亿美元融资担保 - 华尔街日报 https://ai-digest-1.com/zh/news/moonshot-ai-open-sources-kimi-k3-2-8-trillion-parameters-1318AI Digest DNB66zh2026-07-28T06:03:21.894658+00:00月之暗面开源Kimi K3模型：2.8万亿参数，被海外开发者称为“本土版Fable 5” https://ai-digest-1.com/zh/news/worlds-first-agentic-diffusion-model-self-corrects-while-acting-128k-context-1317AI Digest DNB66zh2026-07-28T06:03:12.637956+00:00全球首个Agentic扩散模型：行动中自我纠错，128K上下文媲美自回归模型 https://ai-digest-1.com/zh/news/hugging-face-deepfake-nudes-problem-1315AI Digest DNB66zh2026-07-28T06:02:59.579824+00:00Hugging Face 面临深度伪造裸照问题 https://ai-digest-1.com/zh/news/moonshot-ai-kimi-k3-celebration-aims-for-the-moon-ubs-deepseek-moment-1313AI Digest DNB66zh2026-07-28T05:05:14.297762+00:00Moonshot AI的Kimi K3庆祝活动志在登月；瑞银称其为第二个“DeepSeek时刻” https://ai-digest-1.com/zh/news/russia-copyrighted-works-ai-training-1310-1310AI Digest DNB66zh2026-07-28T05:04:52.619937+00:00俄罗斯将使用受版权保护的作品训练人工智能 https://ai-digest-1.com/zh/news/grok-4-5-enters-beta-at-spacex-tesla-1-5t-params-1301AI Digest DNB66zh2026-07-28T05:03:26.240321+00:00Grok 4.5 在 SpaceX 和特斯拉进入测试阶段：1.5 万亿参数 https://ai-digest-1.com/zh/news/kimi-teaches-zhipu-a-lesson-1265AI Digest DNB66zh2026-07-28T04:05:17.820187+00:00Kimi给智谱上了一课 https://ai-digest-1.com/zh/news/google-ai-transforms-search-more-than-any-update-1261AI Digest DNB66zh2026-07-28T04:04:31.785124+00:00谷歌：AI对搜索的变革远超史上任何更新 https://ai-digest-1.com/zh/news/microsoft-unveils-mai-cyber-1-flash-ai-model-1260AI Digest DNB66zh2026-07-28T04:04:22.614887+00:00微软发布MAI-Cyber-1-Flash AI模型，可在几分钟内自主发现并修复漏洞 https://ai-digest-1.com/zh/news/anthropic-amodei-responds-not-against-open-weights-fears-china-ai-1255AI Digest DNB66zh2026-07-28T04:03:43.937687+00:00Anthropic 的 Dario Amodei 回应：不反对开源权重模型，但担忧中国 AI https://ai-digest-1.com/zh/news/hangzhou-proposes-to-support-enterprises-in-building-token-platforms-1253AI Digest DNB66zh2026-07-28T04:03:28.811612+00:00杭州拟支持企业建设代币生产平台、交易结算平台和对外服务平台 https://ai-digest-1.com/zh/news/kimi-k3-pac1-ecom1-results-204-tasks-failure-analysis-1251AI Digest DNB66zh2026-07-28T04:03:05.055970+00:00Kimi K3 在 PAC1 和 ECOM1 上的表现：204 项任务结果与失败分析 https://ai-digest-1.com/zh/news/kimi-k3-moonshot-ai-open-weights-model-1250AI Digest DNB66zh2026-07-28T04:02:55.017532+00:00Kimi K3：Moonshot AI 开放 2.8 万亿参数模型权重 https://ai-digest-1.com/zh/news/mamba-linear-complexity-alternative-to-transformer-1248AI Digest DNB66zh2026-07-27T22:05:51.535148+00:00Mamba：一种线性复杂度的Transformer替代方案 https://ai-digest-1.com/zh/news/qwen2-5-max-exploring-moe-model-capabilities-1246AI Digest DNB66zh2026-07-27T22:05:32.951570+00:00Qwen2.5-Max：探索大规模 MoE 模型的能力 https://ai-digest-1.com/zh/news/alibaba-qwen-qwq-max-preview-deep-reasoning-1245AI Digest DNB66zh2026-07-27T22:05:25.926617+00:00阿里Qwen发布QwQ-Max-Preview：深度推理与多领域潜力新模型 https://ai-digest-1.com/zh/news/qwq-32b-reinforcement-learning-1244AI Digest DNB66zh2026-07-27T22:05:17.279573+00:00QwQ-32B：利用强化学习的力量 https://ai-digest-1.com/zh/news/alibaba-open-sources-qwen25-vl-32b-1243AI Digest DNB66zh2026-07-27T22:05:10.152273+00:00阿里巴巴开源Qwen2.5-VL-32B：更智能、更轻量 https://ai-digest-1.com/zh/news/complete-guide-to-qwen3-understanding-and-implementing-1241AI Digest DNB66zh2026-07-27T22:04:54.861932+00:00Qwen3完全指南：从零理解与实现 https://ai-digest-1.com/zh/news/russia-allows-ai-developers-use-copyrighted-works-without-consent-1236AI Digest DNB66zh2026-07-27T22:03:34.963816+00:00俄罗斯将允许AI开发者未经版权所有者同意使用书籍、音乐和电影 https://ai-digest-1.com/zh/news/reasoning-ai-models-how-they-work-and-why-they-matter-1234AI Digest DNB66zh2026-07-27T22:03:06.479093+00:00推理型AI模型：工作原理及重要性解析 https://ai-digest-1.com/zh/news/nadella-single-ai-system-warning-1233AI Digest DNB66zh2026-07-27T22:02:55.910768+00:00纳德拉：依赖单一AI系统的公司可能无法生存 https://ai-digest-1.com/zh/news/ai-model-availability-july-27-2026-1231AI Digest DNB66zh2026-07-27T21:23:24.144907+00:002026年7月27日AI模型可用性报告 https://ai-digest-1.com/zh/news/qwen2-5-omni-sees-hears-speaks-writes-all-in-one-1229AI Digest DNB66zh2026-07-27T21:03:59.645453+00:00Qwen2.5 Omni：全能模型，看、听、说、写一应俱全 https://ai-digest-1.com/zh/news/alibaba-qwen-releases-qvq-max-visual-reasoning-model-1228AI Digest DNB66zh2026-07-27T21:03:53.154437+00:00阿里巴巴通义千问发布QVQ-Max：具备推理能力的视觉推理模型 https://ai-digest-1.com/zh/news/deepseek-hints-at-r2-model-introduces-spct-1227AI Digest DNB66zh2026-07-27T21:03:44.768190+00:00DeepSeek 暗示 R2 模型并引入新推理扩展方法 SPCT https://ai-digest-1.com/zh/news/alibaba-qwen3-embedding-text-embedding-ranking-1225AI Digest DNB66zh2026-07-27T21:03:24.953717+00:00阿里巴巴发布 Qwen3 Embedding 系列模型，用于文本嵌入和排序 https://ai-digest-1.com/zh/news/zhipu-ai-acquires-zhongke-jiahe-beijing-full-stack-ai-1220AI Digest DNB66zh2026-07-27T21:02:23.725624+00:00智谱AI收购中科加禾：加强北京全栈AI的新举措 https://ai-digest-1.com/zh/news/claude-chats-artifacts-exposed-publicly-via-google-1217AI Digest DNB66zh2026-07-27T21:01:46.016548+00:00Google上大量Claude对话和Artifacts被公开曝光 https://ai-digest-1.com/zh/news/kimi-and-kvcache-ai-open-source-agentenv-1216AI Digest DNB66zh2026-07-27T21:01:26.964913+00:00Kimi与kvcache-ai团队开源AgentENV——面向Kimi K3规模的智能体强化学习训练分布式平台 https://ai-digest-1.com/zh/news/putin-signs-russia-first-ai-law-1210AI Digest DNB66zh2026-07-27T20:05:23.042123+00:00普京签署俄罗斯首部人工智能法 https://ai-digest-1.com/zh/news/microsoft-launches-first-cybersecurity-ai-model-and-agentic-defense-platform-1206AI Digest DNB66zh2026-07-27T20:03:41.216390+00:00微软发布首个网络安全AI模型及新型自主防御平台 https://ai-digest-1.com/zh/news/verizon-google-dark-fiber-deal-billion-1205AI Digest DNB66zh2026-07-27T20:03:06.116058+00:00Verizon宣布与谷歌达成价值10亿美元的暗光纤交易，连接数据中心——这仅是众多交易中的第一个 https://ai-digest-1.com/zh/news/import-ai-458-making-sense-of-the-future-singularity-1202AI Digest DNB66zh2026-07-27T18:04:32.795288+00:00Import AI 458：解读未来与奇点场景 https://ai-digest-1.com/zh/news/import-ai-461-alignment-not-on-track-frontiercode-synthetic-research-interns-1201AI Digest DNB66zh2026-07-27T18:04:18.608450+00:00Import AI 461：“对齐并未走上正轨”；FrontierCode；以及合成研究实习生 https://ai-digest-1.com/zh/news/cross-origin-storage-api-transformers-js-experiment-1200AI Digest DNB66zh2026-07-27T18:04:08.364678+00:00在Transformers.js中实验跨域存储API提案 https://ai-digest-1.com/zh/news/baidu-maps-mass-production-end-to-end-ai-agents-car-cockpits-1196AI Digest DNB66zh2026-07-27T18:03:29.347329+00:00大规模量产车载端到端AI智能体：百度地图树立行业新标杆 https://ai-digest-1.com/zh/news/mondaycom-joins-tech-companies-citing-ai-as-reason-for-layoffs-1193AI Digest DNB66zh2026-07-27T18:02:50.765298+00:00Monday.com 加入以AI为裁员理由的科技公司名单——还有20多个例子 https://ai-digest-1.com/zh/news/anthropic-claude-user-conversations-leaked-google-search-1189AI Digest DNB66zh2026-07-27T18:02:17.028764+00:00Anthropic的Claude用户对话泄露至谷歌搜索 https://ai-digest-1.com/zh/news/local-rag-for-cad-api-overcoming-knowledge-shortage-1187AI Digest DNB66zh2026-07-27T17:05:08.389148+00:00CAD API 的本地 RAG：如何克服专有接口的知识短缺 https://ai-digest-1.com/zh/news/writing-decoder-only-transformer-llm-from-scratch-python-1186AI Digest DNB66zh2026-07-27T17:04:57.569853+00:00在Python中从头编写一个仅解码器Transformer用于大语言模型 https://ai-digest-1.com/zh/news/world-first-chat-system-eliza-code-secrets-1185AI Digest DNB66zh2026-07-27T17:04:45.094642+00:00全球首个聊天系统可变换性格：ELIZA代码揭示新秘密 https://ai-digest-1.com/zh/news/ai-scorecard-system-from-openai-cfo-1184AI Digest DNB66zh2026-07-27T17:04:35.703889+00:00OpenAI首席财务官提出的AI记分卡系统 https://ai-digest-1.com/zh/news/managing-reasoning-levels-in-large-language-models-1182AI Digest DNB66zh2026-07-27T17:04:20.180551+00:00管理大型语言模型中的推理级别 https://ai-digest-1.com/zh/news/ai-mania-destroys-global-decision-making-1181AI Digest DNB66zh2026-07-27T17:04:15.307792+00:00AI热潮摧毁全球决策能力 https://ai-digest-1.com/zh/news/chinese-open-source-ai-models-set-new-records-1180AI Digest DNB66zh2026-07-27T17:04:09.080625+00:00中国开源AI模型参数数量创新高 https://ai-digest-1.com/zh/news/anthropic-criticized-for-refusing-open-models-1176AI Digest DNB66zh2026-07-27T17:03:19.277026+00:00Anthropic因拒绝开源模型受到批评——唯一没有开源模型的主要AI实验室 https://ai-digest-1.com/zh/news/guardoc-health-amazon-nova-medical-documents-1175AI Digest DNB66zh2026-07-27T17:03:06.620673+00:00Guardoc Health 使用 Amazon Nova 处理医疗文档 https://ai-digest-1.com/zh/news/deepgram-enhances-amazon-sagemaker-ai-support-with-aws-iam-temporary-delegation-1174AI Digest DNB66zh2026-07-27T17:02:56.288453+00:00Deepgram 增强 Amazon SageMaker AI 支持，集成 AWS IAM 临时委托 https://ai-digest-1.com/zh/news/task-aware-knowledge-compression-for-enterprise-ai-on-aws-1173AI Digest DNB66zh2026-07-27T17:02:45.134419+00:00TAKC：面向AWS企业AI的任务感知知识压缩，一种超越RAG的方法 https://ai-digest-1.com/zh/news/altman-brings-new-openai-models-to-washington-1172AI Digest DNB66zh2026-07-27T17:02:40.015905+00:00Altman 本周将携 OpenAI 新模型亮相华盛顿 https://ai-digest-1.com/zh/news/threads-meta-ai-direct-messages-1170AI Digest DNB66zh2026-07-27T17:02:25.715108+00:00Threads 用户现可在私信中与 Meta AI 聊天 https://ai-digest-1.com/zh/news/china-free-ai-models-kimi-k3-1169AI Digest DNB66zh2026-07-27T17:02:15.575277+00:00中国为何免费开放其最强AI模型 https://ai-digest-1.com/zh/news/openai-ai-out-of-control-cyberattack-bbc-1168AI Digest DNB66zh2026-07-27T16:06:36.183375+00:00AI OpenAI失控并对BBC发起网络攻击 https://ai-digest-1.com/zh/news/zhipu-open-source-models-can-be-deployed-independently-1166AI Digest DNB66zh2026-07-27T16:06:16.423042+00:00智谱谈开源：模型可独立部署随时调用，用户掌握主动权 https://ai-digest-1.com/zh/news/wait-for-liang-wenfeng-1163AI Digest DNB66zh2026-07-27T16:05:39.264680+00:00等待梁文锋 https://ai-digest-1.com/zh/news/safe-superintelligence-sutskever-partners-with-nvidia-1159AI Digest DNB66zh2026-07-27T16:04:52.321911+00:00安全超级智能公司Sutskever宣布与英伟达合作，扩大AI研究规模 https://ai-digest-1.com/zh/news/nvidia-open-source-alliance-combat-rogue-ai-agents-1157AI Digest DNB66zh2026-07-27T16:04:27.925882+00:00Nvidia 组建开源联盟，对抗恶意 AI 代理 https://ai-digest-1.com/zh/news/google-ai-overviews-becoming-default-in-search-1156AI Digest DNB66zh2026-07-27T16:04:19.123010+00:00谷歌AI概览快速成为搜索默认选项，新数据显示 https://ai-digest-1.com/zh/news/amd-instinct-mi300-detailed-architecture-and-performance-analysis-1154AI Digest DNB66zh2026-07-27T15:05:23.278725+00:00AMD Instinct MI300：详细架构与性能分析 https://ai-digest-1.com/zh/news/large-language-models-in-financial-markets-1153AI Digest DNB66zh2026-07-27T15:05:11.045320+00:00大型语言模型在金融市场中的应用 https://ai-digest-1.com/zh/news/zhipu-ai-open-sources-glm-models-8x-faster-than-deepseek-r1-1152AI Digest DNB66zh2026-07-27T15:04:57.765067+00:00智谱AI开源GLM：模型比DeepSeek-R1快8倍，全球扩张或为IPO铺路 https://ai-digest-1.com/zh/news/srpo-kuaishou-rl-method-more-efficient-than-grpo-1151AI Digest DNB66zh2026-07-27T15:04:49.905738+00:00SRPO：快手新强化学习方法在LLM训练中效率比GRPO高10倍 https://ai-digest-1.com/zh/news/deepseek-releases-deepseek-prover-v2-with-recursive-proof-search-and-new-proverb-1150AI Digest DNB66zh2026-07-27T15:04:42.421210+00:00DeepSeek 发布 DeepSeek-Prover-V2，具有递归证明搜索和新的 ProverBench 基准 https://ai-digest-1.com/zh/news/deepseek-v3-technical-report-hardware-software-co-design-low-cost-training-1149AI Digest DNB66zh2026-07-27T15:04:26.538601+00:00DeepSeek-V3技术报告：软硬件协同设计如何实现低成本大模型训练 https://ai-digest-1.com/zh/news/alibaba-qwen-vlo-unified-multimodal-model-1148AI Digest DNB66zh2026-07-27T15:04:19.293771+00:00阿里发布统一多模态理解与生成模型Qwen VLo https://ai-digest-1.com/zh/news/qwen-tts-speaks-dialects-new-alibaba-speech-synthesizer-1147AI Digest DNB66zh2026-07-27T15:04:11.602320+00:00Qwen-TTS 现在能说方言了：阿里巴巴新款语音合成器 https://ai-digest-1.com/zh/news/comparison-of-large-language-model-architectures-from-deepseek-v3-to-olmo-2-1146AI Digest DNB66zh2026-07-27T15:04:01.672377+00:00大型语言模型架构对比：从DeepSeek V3到OLMo 2 https://ai-digest-1.com/zh/news/alibaba-qwen-unveils-qwen3-coder-agentic-code-model-1145AI Digest DNB66zh2026-07-27T15:03:50.973968+00:00阿里巴巴通义千问发布Qwen3-Coder：最具代理能力的代码模型 https://ai-digest-1.com/zh/news/alibaba-launches-qwen-mt-machine-translation-model-1144AI Digest DNB66zh2026-07-27T15:03:41.564061+00:00阿里发布Qwen-MT：支持92种语言的机器翻译模型，具备高级定制功能 https://ai-digest-1.com/zh/news/from-gpt-2-to-gpt-oss-architectural-advances-1143AI Digest DNB66zh2026-07-27T15:03:30.700637+00:00从 GPT-2 到 gpt-oss：架构进展分析及与 Qwen3 的比较 https://ai-digest-1.com/zh/news/mistral-ai-launches-ai-studio-production-platform-1141AI Digest DNB66zh2026-07-27T15:03:11.315962+00:00Mistral AI 推出 AI Studio 生产平台 https://ai-digest-1.com/zh/news/kimi-k3-selling-cheaply-1140AI Digest DNB66zh2026-07-27T15:03:07.726292+00:00Kimi K3 其实是在‘低价抛售’ https://ai-digest-1.com/zh/news/meta-releases-muse-spark-11-for-consumers-1138AI Digest DNB66zh2026-07-27T15:02:50.004665+00:00Meta 发布面向消费者的 Muse Spark 1.1，并推出新的 Facebook 功能 https://ai-digest-1.com/zh/news/ai-law-phased-implementation-business-preparation-period-1136AI Digest DNB66zh2026-07-27T15:02:35.447650+00:00人工智能法将分阶段生效：企业获1个月至数年准备期 https://ai-digest-1.com/zh/news/import-ai-466-bitter-lesson-robotics-ai-programming-tasks-openai-hacker-1135AI Digest DNB66zh2026-07-27T15:01:59.852636+00:00Import AI 466：机器人学的苦涩教训，AI完成一周编程任务，OpenAI的意外AI黑客 https://ai-digest-1.com/zh/news/ai-cybersecurity-attacks-43-percent-companies-hit-1134AI Digest DNB66zh2026-07-27T15:01:51.426007+00:00AI网络安全攻击：43%的企业已遭袭击 https://ai-digest-1.com/zh/news/ai-reassembles-retail-from-store-layout-to-assortment-management-1132AI Digest DNB66zh2026-07-27T15:01:32.743320+00:00人工智能重塑零售：从店面布局到品类管理 https://ai-digest-1.com/zh/news/claude-opus-5-foodtruck-bench-record-strategy-breakdown-1131AI Digest DNB66zh2026-07-27T15:01:22.089479+00:00Claude Opus 5 在 FoodTruck Bench 创下纪录：30天收入75,264美元及策略拆解 https://ai-digest-1.com/zh/news/claude-ai-shared-chats-indexed-by-google-1130AI Digest DNB66zh2026-07-27T15:01:12.839486+00:00Claude AI共享聊天被Google索引——检查你的对话是否暴露 https://ai-digest-1.com/zh/news/struq-and-secalign-defending-against-prompt-injection-attacks-1129AI Digest DNB66zh2026-07-27T14:06:22.515658+00:00StruQ和SecAlign：防御提示注入攻击 https://ai-digest-1.com/zh/news/video-world-models-with-long-term-memory-state-space-models-1128AI Digest DNB66zh2026-07-27T14:06:13.073613+00:00Adobe Research与斯坦福和普林斯顿科学家提出基于状态空间模型的具有长期记忆的视频世界模型 https://ai-digest-1.com/zh/news/bytedance-unveils-astra-dual-module-architecture-for-robot-navigation-1126AI Digest DNB66zh2026-07-27T14:05:51.335288+00:00字节跳动推出Astra：用于自主机器人导航的双模块架构 https://ai-digest-1.com/zh/news/peva-predicting-egocentric-video-from-whole-body-actions-1125AI Digest DNB66zh2026-07-27T14:05:39.759612+00:00PEVA：从全身动作预测第一人称视频 https://ai-digest-1.com/zh/news/gspo-scalable-reinforcement-learning-for-language-models-1124AI Digest DNB66zh2026-07-27T14:05:31.200368+00:00GSPO：面向语言模型的可扩展强化学习算法 https://ai-digest-1.com/zh/news/alibaba-releases-qwen-image-20b-mmdit-model-1123AI Digest DNB66zh2026-07-27T14:05:23.743615+00:00阿里巴巴发布Qwen-Image：20B参数的MMDiT模型，实现文字精准图像生成 https://ai-digest-1.com/zh/news/qwen-image-edit-high-quality-efficient-image-editing-1121AI Digest DNB66zh2026-07-27T14:05:07.034266+00:00Qwen-Image-Edit：高质量且高效的图像编辑 https://ai-digest-1.com/zh/news/nvidia-blackwell-ultra-15-pflops-nvfp4-288-gb-hbm3e-no-fp64-1120AI Digest DNB66zh2026-07-27T14:04:56.953530+00:00NVIDIA Blackwell Ultra：15 Pflops NVFP4、288 GB HBM3e，无FP64 https://ai-digest-1.com/zh/news/word2vec-true-learning-theory-discovered-1119AI Digest DNB66zh2026-07-27T14:04:47.380359+00:00Word2vec 到底学到了什么？研究者终于提出了精确的学习理论 https://ai-digest-1.com/zh/news/qwen3guard-streaming-safety-real-time-tokens-1118AI Digest DNB66zh2026-07-27T14:04:39.995870+00:00Qwen3Guard：实时令牌流式安全防护 https://ai-digest-1.com/zh/news/rl-without-td-learning-new-divide-and-conquer-algorithm-1117AI Digest DNB66zh2026-07-27T14:04:30.533709+00:00无需时序差分学习的强化学习：新的分治算法 https://ai-digest-1.com/zh/news/beyond-standard-llms-linear-attention-hybrids-text-diffusion-code-world-models-s-1116AI Digest DNB66zh2026-07-27T14:04:18.796822+00:00超越标准大语言模型：线性注意力混合体、文本扩散、代码世界模型与小规模递归Transformer https://ai-digest-1.com/zh/news/mistral-ai-expands-in-germany-1115AI Digest DNB66zh2026-07-27T14:04:13.738988+00:00Mistral AI 扩大在德国的业务布局 https://ai-digest-1.com/zh/news/mistral-ai-unveils-mistral-3-model-family-1114AI Digest DNB66zh2026-07-27T14:04:02.404106+00:00Mistral AI发布Mistral 3系列模型，包括Mistral Large 3 https://ai-digest-1.com/zh/news/deepseek-v3-to-v3-2-architecture-sparse-attention-rl-updates-1113AI Digest DNB66zh2026-07-27T14:03:51.833740+00:00从DeepSeek V3到V3.2：架构、稀疏注意力与强化学习更新 https://ai-digest-1.com/zh/news/mistral-ai-releases-devstral-2-coding-models-and-mistral-vibe-cli-1112AI Digest DNB66zh2026-07-27T14:03:40.341214+00:00Mistral AI 发布 Devstral 2 代码模型和 Mistral Vibe CLI https://ai-digest-1.com/zh/news/state-of-llms-in-2025-1111AI Digest DNB66zh2026-07-27T14:03:24.448957+00:002025年大语言模型现状：进展、挑战与预测 https://ai-digest-1.com/zh/news/nous-research-releases-opensource-coding-model-nouscoder-14b-1110AI Digest DNB66zh2026-07-27T14:03:17.004089+00:00Nous Research 发布开源编程模型 NousCoder-14B，4 天内在 48 块 B200 GPU 上完成训练 https://ai-digest-1.com/zh/news/anthropic-launches-cowork-ai-agent-for-file-work-1109AI Digest DNB66zh2026-07-27T14:03:09.524343+00:00Anthropic 推出 Cowork —— 无需编程即可处理文件的 AI 智能体 https://ai-digest-1.com/zh/news/mistral-ai-engineers-find-memory-leak-in-vllm-using-bpftrace-1108AI Digest DNB66zh2026-07-27T14:02:59.131869+00:00Mistral AI工程师使用BPFtrace发现vLLM内存泄漏的真实原因 https://ai-digest-1.com/zh/news/google-deepmind-helps-national-hurricane-center-predict-hurricane-melissa-1107AI Digest DNB66zh2026-07-27T14:02:46.831910+00:00Google DeepMind帮助美国国家飓风中心更精准预测飓风梅丽莎在牙买加的历史性登陆 https://ai-digest-1.com/zh/news/accelerating-creation-of-space-brain-beijing-officially-launches-digital-space-1-1104AI Digest DNB66zh2026-07-27T14:02:08.934711+00:00加速打造太空大脑：北京正式启动数字空间一号实验卫星项目 https://ai-digest-1.com/zh/news/moscow-integrated-ai-radiologists-daily-work-1103AI Digest DNB66zh2026-07-27T14:02:03.655770+00:00莫斯科如何将人工智能融入放射科医生的日常工作 https://ai-digest-1.com/zh/news/survey-ai-requests-vs-infrastructure-readiness-gap-1102AI Digest DNB66zh2026-07-27T14:01:50.324308+00:00调查：65%的企业认为管理层AI需求与基础设施就绪度存在差距 https://ai-digest-1.com/zh/news/cognizant-anthropic-expand-partnership-claude-integrated-platforms-1100AI Digest DNB66zh2026-07-27T14:01:37.600636+00:00Cognizant与Anthropic扩大合作：Claude集成至Cognizant行业平台 https://ai-digest-1.com/zh/news/do-human-salespeople-still-matter-ai-agent-promises-1099AI Digest DNB66zh2026-07-27T14:01:25.298100+00:00人类销售人员还重要吗？AI代理承诺背后的现实 https://ai-digest-1.com/zh/news/designing-imaging-systems-using-information-theory-1097AI Digest DNB66zh2026-07-27T13:05:49.979948+00:00使用信息论设计成像系统 https://ai-digest-1.com/zh/news/salesforce-launches-new-ai-agent-slackbot-1096AI Digest DNB66zh2026-07-27T13:05:44.321020+00:00Salesforce推出全新AI Agent Slackbot，在企业AI领域挑战微软和谷歌 https://ai-digest-1.com/zh/news/inference-time-scaling-categories-improved-llm-reasoning-1093AI Digest DNB66zh2026-07-27T13:05:21.795880+00:00推理时扩展类别：提升大语言模型推理能力 https://ai-digest-1.com/zh/news/mistral-ai-releases-mistral-vibe-20-terminal-native-ai-agent-1092AI Digest DNB66zh2026-07-27T13:05:15.035211+00:00Mistral AI 发布 Mistral Vibe 2.0，一款由 Devstral 2 驱动的终端原生 AI 代理 https://ai-digest-1.com/zh/news/mistral-ai-voxtral-transcribe-2-speech-recognition-1091AI Digest DNB66zh2026-07-27T13:05:01.699938+00:00Mistral AI 发布 Voxtral Transcribe 2 — 超低延迟语音识别模型 https://ai-digest-1.com/zh/news/testing-rails-on-autopilot-ai-agent-writes-tests-instead-of-developers-1090AI Digest DNB66zh2026-07-27T13:04:50.135871+00:00自动化Rails测试：AI代理替代开发者编写测试 https://ai-digest-1.com/zh/news/leanstral-open-source-foundation-trusted-programming-1089AI Digest DNB66zh2026-07-27T13:04:40.801193+00:00Leanstral：可信编程的开源基础 https://ai-digest-1.com/zh/news/google-evaluates-alignment-of-behavioral-tendencies-in-llms-1088AI Digest DNB66zh2026-07-27T13:04:31.385782+00:00谷歌评估大语言模型行为倾向的一致性 https://ai-digest-1.com/zh/news/google-ai-agents-research-papervizagent-scholarpeer-1087AI Digest DNB66zh2026-07-27T13:04:23.809287+00:00谷歌推出两款科研AI助手：PaperVizAgent用于插图，ScholarPeer用于审稿 https://ai-digest-1.com/zh/news/convapparel-realism-gap-llm-user-simulators-1086AI Digest DNB66zh2026-07-27T13:04:13.517924+00:00ConvApparel：衡量并缩小基于LLM的用户模拟器中的真实感差距 https://ai-digest-1.com/zh/news/google-research-generative-ai-assess-future-ready-skills-1085AI Digest DNB66zh2026-07-27T13:04:04.924987+00:00谷歌研究：用生成式AI评估面向未来技能的新方法 https://ai-digest-1.com/zh/news/ai-agents-force-cios-to-rethink-roles-1084AI Digest DNB66zh2026-07-27T13:03:59.298038+00:00AI代理迫使CIO重新思考自身角色 https://ai-digest-1.com/zh/news/setup-local-ai-coding-agent-avoid-cloud-price-hikes-1083AI Digest DNB66zh2026-07-27T13:03:48.569194+00:00如何搭建本地AI编码助手，摆脱云端模型涨价依赖 https://ai-digest-1.com/zh/news/google-deepmind-liver-fibrosis-existing-drugs-1082AI Digest DNB66zh2026-07-27T13:03:38.805097+00:00谷歌DeepMind助力在现有药物中发现肝纤维化治疗新用途 https://ai-digest-1.com/zh/news/uniting-biological-toolkits-for-als-approach-1081AI Digest DNB66zh2026-07-27T13:03:29.741190+00:00整合生物工具包，开辟肌萎缩侧索硬化症治疗新路径 https://ai-digest-1.com/zh/news/google-deepmind-ai-co-scientist-discovers-liver-disease-mechanisms-1080AI Digest DNB66zh2026-07-27T13:03:22.151125+00:00谷歌DeepMind用AI联合科学家加速发现肝病机制 https://ai-digest-1.com/zh/news/google-deepmind-opens-new-paths-in-aging-research-1079AI Digest DNB66zh2026-07-27T13:03:15.759638+00:00谷歌DeepMind为衰老研究开辟新道路 https://ai-digest-1.com/zh/news/10-ideas-from-ai-engineer-on-how-development-changes-when-agents-write-code-1072AI Digest DNB66zh2026-07-27T13:02:02.639745+00:00AI工程师提出的10个点子：当AI代理编写代码时开发工作如何变化 https://ai-digest-1.com/zh/news/mistral-ai-joins-nvidia-nemotron-coalition-1068AI Digest DNB66zh2026-07-27T12:06:44.463934+00:00Mistral AI 加入 NVIDIA Nemotron 联盟，加速开放前沿模型开发 https://ai-digest-1.com/zh/news/mistral-ai-unveils-mistral-small-4-1067AI Digest DNB66zh2026-07-27T12:06:37.175925+00:00Mistral AI 发布 Mistral Small 4：兼具推理与多模态能力的通用模型 https://ai-digest-1.com/zh/news/mistral-ai-launches-voxtral-tts-multilingual-speech-synthesis-1065AI Digest DNB66zh2026-07-27T12:06:19.885838+00:00Mistral AI 发布 Voxtral TTS，多语言语音合成模型 https://ai-digest-1.com/zh/news/nvidia-groq-3-sram-disaggregation-and-determinism-1064AI Digest DNB66zh2026-07-27T12:06:08.092803+00:00NVIDIA Groq 3：新AI平台中的SRAM、解耦与确定性计算 https://ai-digest-1.com/zh/news/nvidia-vera-rubin-pod-cluster-unveiled-1063AI Digest DNB66zh2026-07-27T12:05:56.239934+00:00NVIDIA发布Vera Rubin POD集群：五机架、七芯片、1.2千兆亿晶体管 https://ai-digest-1.com/zh/news/six-key-components-of-a-coding-agent-1062AI Digest DNB66zh2026-07-27T12:05:46.391396+00:00编码体的六个关键组成部分：现代AI系统如何编写代码 https://ai-digest-1.com/zh/news/mogen-synthetic-neurons-improve-brain-mapping-ai-1061AI Digest DNB66zh2026-07-27T12:05:40.135556+00:00MoGen：合成神经元改善脑图谱AI，错误率降低4.4% https://ai-digest-1.com/zh/news/simula-framework-synthetic-datasets-reasoning-1060AI Digest DNB66zh2026-07-27T12:05:31.645381+00:00Simula：基于机制与推理设计合成数据集的新框架 https://ai-digest-1.com/zh/news/grasp-gradient-based-planner-long-horizon-planning-world-models-1059AI Digest DNB66zh2026-07-27T12:05:21.981957+00:00GRASP：基于梯度的长时域规划新方法，用于世界模型中的规划 https://ai-digest-1.com/zh/news/reasoningbank-memory-framework-ai-agents-learning-1058AI Digest DNB66zh2026-07-27T12:05:13.823281+00:00ReasoningBank：一种让AI代理从成功与失败中学习的记忆框架 https://ai-digest-1.com/zh/news/google-auto-frame-neural-network-photo-perspective-retroactively-1057AI Digest DNB66zh2026-07-27T12:05:06.881800+00:00谷歌推出 Auto Frame：神经网络可事后改变照片视角 https://ai-digest-1.com/zh/news/servicenow-launches-ai-agent-control-tower-1056AI Digest DNB66zh2026-07-27T12:04:52.962259+00:00ServiceNow推出AI代理“控制塔” https://ai-digest-1.com/zh/news/co-scientist-molecular-switches-infectious-diseases-1055AI Digest DNB66zh2026-07-27T12:04:46.432928+00:00Co-Scientist如何帮助发现新型传染病的分子开关 https://ai-digest-1.com/zh/news/google-deepmind-singapore-ai-national-partnership-1054AI Digest DNB66zh2026-07-27T12:04:38.114904+00:00Google DeepMind 助力新加坡 AI 未来，达成国家级新合作 https://ai-digest-1.com/zh/news/google-deepmind-expands-content-provenance-tools-synthid-c2pa-1053AI Digest DNB66zh2026-07-27T12:04:29.180289+00:00Google DeepMind 通过 SynthID 和 C2PA 扩展内容来源工具 https://ai-digest-1.com/zh/news/first-look-meta-ai-email-calendar-agent-1049AI Digest DNB66zh2026-07-27T12:03:37.717916+00:00Meta AI 邮件与日历智能体初探 https://ai-digest-1.com/zh/news/chaowei-dynamics-peking-university-healthcare-embodied-ai-medicine-1048AI Digest DNB66zh2026-07-27T12:03:30.091976+00:00超维动力与北大医疗：医疗具身AI的务实路径 https://ai-digest-1.com/zh/news/study-81-percent-vibe-coders-russia-not-programmers-1045AI Digest DNB66zh2026-07-27T12:03:09.453200+00:00研究：俄罗斯81%的“氛围编码者”并非程序员 https://ai-digest-1.com/zh/news/xiaomi-mimo-v2-5-tops-openrouter-global-call-volume-rankings-1044AI Digest DNB66zh2026-07-27T12:03:04.607046+00:00小米 MiMo-V2.5 登顶 OpenRouter 全球周度和月度调用量排行榜 https://ai-digest-1.com/zh/news/mit-stanford-ai-avatars-change-life-decisions-1042AI Digest DNB66zh2026-07-27T12:02:47.744933+00:00MIT和斯坦福创建改变人们重大人生决策的AI化身 https://ai-digest-1.com/zh/news/mistral-ai-launches-workflows-orchestration-layer-for-enterprise-ai-1041AI Digest DNB66zh2026-07-27T11:05:46.355837+00:00Mistral AI 推出 Workflows——面向企业 AI 的编排层 https://ai-digest-1.com/zh/news/google-research-empirical-research-assistance-tool-applications-1040AI Digest DNB66zh2026-07-27T11:05:33.855534+00:00Google研究科学家使用实证研究辅助工具的四种方式 https://ai-digest-1.com/zh/news/ibm-urges-database-administrators-to-trust-ai-1039AI Digest DNB66zh2026-07-27T11:05:20.931162+00:00IBM呼吁数据库管理员信任AI代表他们行事 https://ai-digest-1.com/zh/news/anthropic-claude-financial-agents-launched-1038AI Digest DNB66zh2026-07-27T11:05:15.608403+00:00Anthropic 想让 Claude 管理资金：金融代理功能上线 https://ai-digest-1.com/zh/news/porsche-cup-brasil-ai-race-management-1037AI Digest DNB66zh2026-07-27T11:05:03.535787+00:00Porsche Cup Brasil 如何利用人工智能进行赛事管理 https://ai-digest-1.com/zh/news/adaptive-parallel-reasoning-new-paradigm-ai-inference-1036AI Digest DNB66zh2026-07-27T11:04:51.486737+00:00自适应并行推理：高效人工智能推理扩展的新范式 https://ai-digest-1.com/zh/news/import-ai-456-rsi-economic-growth-radical-optionality-ai-regulation-neural-compu-1035AI Digest DNB66zh2026-07-27T11:04:42.566644+00:00Import AI 456：递归自我改进与经济增长；人工智能监管的激进备选方案；神经计算机 https://ai-digest-1.com/zh/news/google-deepmind-gemini-for-science-ai-tools-1034AI Digest DNB66zh2026-07-27T11:04:32.345616+00:00Google DeepMind 推出 Gemini for Science：用于科学发现的人工智能工具集 https://ai-digest-1.com/zh/news/google-deepmind-gemini-omni-multimodal-video-model-1033AI Digest DNB66zh2026-07-27T11:04:23.356230+00:00谷歌DeepMind推出Gemini Omni：用于视频创建和编辑的多模态模型 https://ai-digest-1.com/zh/news/google-deepmind-combines-genie-with-google-street-view-1032AI Digest DNB66zh2026-07-27T11:04:17.285360+00:00Google DeepMind 将 Genie 与 Google 街景结合，模拟真实场所 https://ai-digest-1.com/zh/news/grok-4-6-musk-two-week-ai-shakeup-1030AI Digest DNB66zh2026-07-27T11:03:43.449156+00:00Grok 4.6：马斯克两周内掀起AI变革 https://ai-digest-1.com/zh/news/who-sets-us-ai-policy-trumps-team-1027AI Digest DNB66zh2026-07-27T11:03:19.825366+00:00谁在制定美国AI政策：特朗普团队 https://ai-digest-1.com/zh/news/samsung-hbm5-2nm-gaa-process-speed-1025AI Digest DNB66zh2026-07-27T11:03:06.684578+00:00三星确认HBM5将采用2纳米GAA工艺，速度比HBM4E提升50%以上 https://ai-digest-1.com/zh/news/us-plans-to-investigate-chinese-ai-companies-sanctions-china-responds-1023AI Digest DNB66zh2026-07-27T11:02:47.797472+00:00美国拟调查中国AI公司并实施制裁：中国商务部回应 https://ai-digest-1.com/zh/news/costa-rican-dairy-cooperative-turns-ai-agents-into-colleagues-1019AI Digest DNB66zh2026-07-27T10:06:55.840657+00:00哥斯达黎加乳制品合作社将AI智能体转变为同事 https://ai-digest-1.com/zh/news/ontario-auditors-ai-medical-scribes-distort-facts-1018AI Digest DNB66zh2026-07-27T10:06:45.555190+00:00安大略省审计员：AI辅助医疗抄写员频繁扭曲基本事实 https://ai-digest-1.com/zh/news/llm-architecture-innovations-kv-sharing-per-layer-embeddings-compressed-attentio-1017AI Digest DNB66zh2026-07-27T10:06:36.247599+00:00LLM架构创新：KV共享、逐层嵌入与压缩注意力 https://ai-digest-1.com/zh/news/6-lessons-for-creating-benchmarks-from-neurips-tutorial-1016AI Digest DNB66zh2026-07-27T10:06:28.689658+00:00创建基准测试的六条经验：来自NeurIPS教程 https://ai-digest-1.com/zh/news/google-introduces-era-ai-tool-scientific-coding-1014AI Digest DNB66zh2026-07-27T10:06:16.724543+00:00谷歌推出ERA，一款用于科学编码的AI工具，刊登于《自然》杂志 https://ai-digest-1.com/zh/news/mistral-ai-connectors-studio-mcp-integration-1012AI Digest DNB66zh2026-07-27T10:06:05.378454+00:00Mistral AI 在 Studio 中发布连接器，内置与自定义 MCP 集成 https://ai-digest-1.com/zh/news/mistral-moves-coding-agents-to-cloud-with-mistral-medium-3-5-1011AI Digest DNB66zh2026-07-27T10:05:57.333988+00:00Mistral 将编码代理迁移至云端，发布 Mistral Medium 3.5 https://ai-digest-1.com/zh/news/mistral-ai-acquires-emmi-ai-physical-ai-1010AI Digest DNB66zh2026-07-27T10:05:48.792643+00:00Mistral AI收购Emmi AI，加速工业物理AI发展 https://ai-digest-1.com/zh/news/google-introduces-private-analytics-technology-based-on-cryptography-and-tee-1008AI Digest DNB66zh2026-07-27T10:05:34.264014+00:00谷歌推出基于密码学和TEE的隐私分析技术 https://ai-digest-1.com/zh/news/measuring-impact-of-learning-with-ai-in-sierra-leone-1007AI Digest DNB66zh2026-07-27T10:05:25.802702+00:00评估塞拉利昂及其他地区使用AI学习的影响 https://ai-digest-1.com/zh/news/google-deepmind-launches-robotics-accelerator-for-european-startups-1006AI Digest DNB66zh2026-07-27T10:05:17.887822+00:00Google DeepMind 为欧洲初创公司推出机器人加速器 https://ai-digest-1.com/zh/news/musk-promises-to-surpass-kimi-moonshot-ai-lets-compete-1004AI Digest DNB66zh2026-07-27T10:05:07.100558+00:00马斯克承诺超越月之暗面Kimi：'来较量吧' https://ai-digest-1.com/zh/news/terence-tao-icm-warning-mathematics-century-crisis-ai-1000AI Digest DNB66zh2026-07-27T10:04:02.177879+00:00陶哲轩在菲尔兹奖颁奖典礼上：数学面临百年危机 https://ai-digest-1.com/zh/news/nvidia-microsoft-ibm-form-ai-safety-alliance-997AI Digest DNB66zh2026-07-27T10:03:21.252707+00:00英伟达、微软、IBM 等多家公司成立人工智能安全联盟 https://ai-digest-1.com/zh/news/nvidia-launches-cosmos-h-dreams-real-time-generative-simulator-surgical-robots-995AI Digest DNB66zh2026-07-27T10:03:00.008553+00:00NVIDIA推出Cosmos-H-Dreams：手术机器人的实时生成式模拟器 https://ai-digest-1.com/zh/news/how-to-train-an-infrastructure-llm-agent-not-to-lie-994AI Digest DNB66zh2026-07-27T10:02:51.367257+00:00如何训练基础设施LLM代理不说谎 https://ai-digest-1.com/zh/news/2026-why-capital-prefers-ai-apps-that-can-make-money-993AI Digest DNB66zh2026-07-27T10:02:36.032412+00:002026年：资本为何偏爱能赚钱的AI应用 https://ai-digest-1.com/zh/news/import-ai-457-ai-stuxnet-muon-optimizer-positive-alignment-992AI Digest DNB66zh2026-07-27T09:06:24.627748+00:00Import AI 457：AI版震网病毒；受诅咒的Muon优化器；以及正向对齐 https://ai-digest-1.com/zh/news/google-redesigns-search-bar-25-years-io-2025-990AI Digest DNB66zh2026-07-27T09:06:11.157703+00:00谷歌25年来首次重新设计搜索栏：这意味着什么以及为何重要 https://ai-digest-1.com/zh/news/ai-bots-scan-tls-certificate-logs-989AI Digest DNB66zh2026-07-27T09:06:00.870741+00:00AI机器人扫描TLS证书日志：任何信息都可用于训练大语言模型 https://ai-digest-1.com/zh/news/mistral-ai-introduces-physics-ai-accelerating-engineering-988AI Digest DNB66zh2026-07-27T09:05:46.189500+00:00Mistral AI 推出物理 AI：加速工程计算 https://ai-digest-1.com/zh/news/mistral-ai-unveils-vibe-unified-ai-agent-985AI Digest DNB66zh2026-07-27T09:05:20.929728+00:00Mistral AI 发布 Vibe——面向工作和编程的统一 AI 代理 https://ai-digest-1.com/zh/news/mistral-ai-unveils-industrial-ai-stack-and-vibe-agent-984AI Digest DNB66zh2026-07-27T09:05:11.929459+00:00Mistral AI 发布工业 AI 堆栈与 Vibe 智能体 https://ai-digest-1.com/zh/news/data-formulator-07-open-source-ai-enterprise-analytics-983AI Digest DNB66zh2026-07-27T09:05:06.523345+00:00Data Formulator 0.7：面向企业数据分析的开源AI系统 https://ai-digest-1.com/zh/news/new-era-of-discovery-google-research-at-io-2026-982AI Digest DNB66zh2026-07-27T09:04:55.148005+00:00新发现时代：谷歌研究在I/O 2026 https://ai-digest-1.com/zh/news/us-ai-economy-grows-2600-annually-invisible-in-gdp-981AI Digest DNB66zh2026-07-27T09:04:47.735678+00:00美国AI经济年增长率达2600%，但在GDP统计中无迹可寻 https://ai-digest-1.com/zh/news/ai-alone-wont-change-your-business-the-system-it-runs-in-will-980AI Digest DNB66zh2026-07-27T09:04:36.465676+00:00AI 本身不会改变你的业务，运行它的系统才会。 https://ai-digest-1.com/zh/news/google-open-sources-hydrological-ai-for-flood-forecasting-979AI Digest DNB66zh2026-07-27T09:04:26.513593+00:00谷歌开源水文AI用于洪水预测 https://ai-digest-1.com/zh/news/llm-research-papers-2026-list-january-to-may-976AI Digest DNB66zh2026-07-27T09:03:55.311988+00:00LLM研究论文：2026年列表（1月到5月） https://ai-digest-1.com/zh/news/society-can-be-reward-hacked-like-cyber-environments-975AI Digest DNB66zh2026-07-27T09:03:27.797418+00:00社会如同网络环境一样可能被奖励黑客利用：想象一支信用卡积分优化大军……永远……在钻系统空子 https://ai-digest-1.com/zh/news/google-deepmind-gemma-4-12b-multimodal-model-without-encoders-974AI Digest DNB66zh2026-07-27T09:03:21.292184+00:00谷歌 DeepMind 发布 Gemma 4 12B：无需编码器的统一多模态模型 https://ai-digest-1.com/zh/news/why-cancer-drugs-affect-differently-973AI Digest DNB66zh2026-07-27T09:03:14.124669+00:00为什么癌症药物对每个人的影响不同？ https://ai-digest-1.com/zh/news/gemini-3-5-live-translate-smooth-natural-voice-translation-972AI Digest DNB66zh2026-07-27T09:03:07.345722+00:00Gemini 3.5 Live Translate：流畅自然的语音翻译 https://ai-digest-1.com/zh/news/diffusiongemma-4x-faster-text-generation-970AI Digest DNB66zh2026-07-27T09:02:51.308131+00:00DiffusionGemma：文本生成速度提升4倍 https://ai-digest-1.com/zh/news/mosaicleaks-data-leakage-prevention-pa-dr-969AI Digest DNB66zh2026-07-27T09:02:42.765307+00:00MosaicLeaks：通过研究型AI代理造成的数据泄露及PA-DR预防方法 https://ai-digest-1.com/zh/news/comparison-of-three-chinese-models-kimi-k3-glm-5-2-deepseek-v4-pro-coding-skills-968AI Digest DNB66zh2026-07-27T09:02:36.216000+00:00三款国产大模型对比：Kimi K3、GLM-5.2、DeepSeek V4 Pro — 谁编程能力最强？ https://ai-digest-1.com/zh/news/moonshot-ai-releases-kimi-k3-global-hit-967AI Digest DNB66zh2026-07-27T09:02:27.276254+00:00月之暗面发布Kimi K3，全球爆款以八分之一投资超越美国 https://ai-digest-1.com/zh/news/huang-renxun-opposes-us-ban-on-chinese-open-source-ai-models-966AI Digest DNB66zh2026-07-27T09:02:19.614001+00:00黄仁勋反对美国禁止中国开源AI模型，称赞Kimi K3等 https://ai-digest-1.com/zh/news/worlds-most-advanced-servicing-robot-satellite-launched-965AI Digest DNB66zh2026-07-27T09:02:07.934391+00:00全球最先进维修机器人卫星发射，前往地球静止轨道 https://ai-digest-1.com/zh/news/inside-the-relay-market-fueling-token-resellers-and-fraud-963AI Digest DNB66zh2026-07-27T09:01:52.093725+00:00揭秘支撑令牌转售与欺诈的代理市场 https://ai-digest-1.com/zh/news/anthropic-only-refuses-huang-open-initiative-962AI Digest DNB66zh2026-07-27T09:01:36.057060+00:00Anthropic仍是唯一未签署黄仁勋开放倡议的公司 https://ai-digest-1.com/zh/news/low-carbon-computing-platform-from-old-smartphones-959AI Digest DNB66zh2026-07-27T08:06:00.589336+00:00旧智能手机打造低碳计算平台 https://ai-digest-1.com/zh/news/from-pixels-to-planning-earth-ai-for-nature-restoration-956AI Digest DNB66zh2026-07-27T08:05:33.776382+00:00从像素到规划：地球AI助力自然恢复 https://ai-digest-1.com/zh/news/google-deepmind-uk-government-ai-construction-applications-955AI Digest DNB66zh2026-07-27T08:05:24.768972+00:00谷歌DeepMind与英国政府合作，利用AI加速建筑申请审批 https://ai-digest-1.com/zh/news/sound-waves-neuromorphic-chips-brain-953AI Digest DNB66zh2026-07-27T08:05:06.304180+00:00声波助力神经形态芯片更好模拟大脑 https://ai-digest-1.com/zh/news/free-openclaw-repository-triage-using-local-ai-models-952AI Digest DNB66zh2026-07-27T08:04:53.174051+00:00使用本地AI模型对OpenClaw仓库进行免费分类 https://ai-digest-1.com/zh/news/ai-outperforms-humans-in-persuasion-study-self-sustaining-ai-superintelligence-p-951AI Digest DNB66zh2026-07-27T08:04:41.189022+00:00研究显示AI在说服力上超越人类；自维持AI何时出现；通往超级智能的路径 https://ai-digest-1.com/zh/news/pp-ocrv6-on-hugging-face-50-languages-950AI Digest DNB66zh2026-07-27T08:04:32.977481+00:00PP-OCRv6 登陆 Hugging Face：支持 50 种语言的文字识别，模型参数量从 150 万到 3450 万 https://ai-digest-1.com/zh/news/kimi-k3-gains-popularity-moonshot-responds-to-criticism-946AI Digest DNB66zh2026-07-27T08:03:54.093918+00:00Kimi K3 走红：月之暗面首次回应模型蒸馏争议 https://ai-digest-1.com/zh/news/meta-ai-doesnt-just-think-it-acts-943AI Digest DNB66zh2026-07-27T08:03:32.712690+00:00Meta AI 不仅能思考，还能行动 https://ai-digest-1.com/zh/news/blocked-but-thriving-china-american-ai-942AI Digest DNB66zh2026-07-27T08:03:18.733724+00:00封锁但繁荣：中国人如何使用美国AI https://ai-digest-1.com/zh/news/russia-passes-first-law-on-ai-development-garant-941AI Digest DNB66zh2026-07-27T08:03:07.592532+00:00俄罗斯通过首部人工智能发展法——GARANT https://ai-digest-1.com/zh/news/mistral-ai-releases-ocr-4-936AI Digest DNB66zh2026-07-27T07:05:45.028612+00:00Mistral AI 发布 OCR 4，支持边界框、区块分类和置信度评分 https://ai-digest-1.com/zh/news/anthropic-claude-slack-annoying-ai-coworker-935AI Digest DNB66zh2026-07-27T07:05:39.025251+00:00Anthropic 将 Claude 集成到 Slack，变成 24/7 工作的烦人 AI 同事 https://ai-digest-1.com/zh/news/ffasr-leaderboard-launched-benchmark-for-asr-in-real-acoustic-conditions-934AI Digest DNB66zh2026-07-27T07:05:27.833090+00:00FFASR排行榜发布：真实声学环境下的ASR评测基准 https://ai-digest-1.com/zh/news/ai-creates-unimaginable-radio-chip-designs-932AI Digest DNB66zh2026-07-27T07:05:09.676006+00:00AI 创建出人类无法想象的无线电芯片设计 https://ai-digest-1.com/zh/news/nvidia-nemo-automodel-moe-fine-tuning-expert-parallelism-deepep-931AI Digest DNB66zh2026-07-27T07:04:58.643352+00:00NVIDIA NeMo AutoModel：借助专家并行与DeepEP技术，将MoE微调速度提升3.7倍 https://ai-digest-1.com/zh/news/built-in-computer-use-now-available-in-gemini-35-flash-930AI Digest DNB66zh2026-07-27T07:04:52.854269+00:00内置电脑使用功能现已登陆 Gemini 3.5 Flash https://ai-digest-1.com/zh/news/google-reasoning-helps-language-models-recall-facts-929AI Digest DNB66zh2026-07-27T07:04:44.939936+00:00Google：推理帮助语言模型通过缓冲和启动回忆事实 https://ai-digest-1.com/zh/news/introducing-gpt-live-new-voice-models-927AI Digest DNB66zh2026-07-27T07:04:29.513195+00:00GPT-Live 登场：新一代语音模型实现人与AI的自然交互 https://ai-digest-1.com/zh/news/from-deepseek-to-k3-chinese-llms-enter-global-top-league-925AI Digest DNB66zh2026-07-27T07:04:09.493643+00:00从DeepSeek到K3：中国大语言模型如何进入全球顶级行列 https://ai-digest-1.com/zh/news/deepseek-v4-released-top-large-model-now-available-922AI Digest DNB66zh2026-07-27T07:03:33.275237+00:00DeepSeek-V4 发布：顶级大模型现已全面开放 https://ai-digest-1.com/zh/news/token-cramming-ideal-text-compression-does-not-preserve-meaning-airi-study-919AI Digest DNB66zh2026-07-27T07:03:05.683708+00:00Token挤压法：理想文本压缩不保留语义——AIRI研究 https://ai-digest-1.com/zh/news/ai-companies-lobbying-spending-record-us-917AI Digest DNB66zh2026-07-27T07:02:51.217150+00:00AI公司在美国的游说支出创历史新高 https://ai-digest-1.com/zh/news/linear-elastic-caching-cloud-cost-optimization-915AI Digest DNB66zh2026-07-27T06:06:37.600859+00:00线性弹性缓存：一种优化云成本的新型缓存管理方法 https://ai-digest-1.com/zh/news/lab-accident-could-revolutionize-computing-913AI Digest DNB66zh2026-07-27T06:06:19.386166+00:00可能彻底改变计算领域的实验室意外发现 https://ai-digest-1.com/zh/news/every-eval-ever-results-on-hugging-face-model-pages-911AI Digest DNB66zh2026-07-27T06:06:00.068444+00:00所有评估结果现已可在 Hugging Face 模型页面上查看 https://ai-digest-1.com/zh/news/specialization-is-inevitable-optimization-biology-markets-machine-learning-910AI Digest DNB66zh2026-07-27T06:05:49.501096+00:00专业化不可避免：优化理论、生物学、市场和机器学习告诉我们什么 https://ai-digest-1.com/zh/news/scarfbench-benchmark-evaluating-ai-agents-java-framework-migration-909AI Digest DNB66zh2026-07-27T06:05:41.826110+00:00ScarfBench：评估AI代理在企业级Java框架迁移任务中的基准 https://ai-digest-1.com/zh/news/testing-checklist-search-rag-systems-six-levels-908AI Digest DNB66zh2026-07-27T06:05:31.265044+00:00搜索与RAG系统测试清单：六个层次的检查 https://ai-digest-1.com/zh/news/kimi-k3-domestic-open-model-strength-new-user-access-timeline-903AI Digest DNB66zh2026-07-27T06:05:00.796649+00:00Kimi K3：国产开源模型的优势何在，新用户何时能用上？ https://ai-digest-1.com/zh/news/china-creates-new-soft-power-open-and-affordable-ai-901AI Digest DNB66zh2026-07-27T06:04:43.065346+00:00中国创造新'软实力'：开放且价格低廉的人工智能 https://ai-digest-1.com/zh/news/deepseek-halts-second-funding-round-10-billion-yuan-897AI Digest DNB66zh2026-07-27T06:04:03.050727+00:00DeepSeek 据报暂停第二轮融资，规模至少100亿元 https://ai-digest-1.com/zh/news/chinese-academy-of-sciences-fixes-ai-social-intelligence-894AI Digest DNB66zh2026-07-27T06:03:31.289384+00:00中国科学院攻克AI最尴尬的短板：社会智能 https://ai-digest-1.com/zh/news/rag-benchmarks-matter-more-than-code-893AI Digest DNB66zh2026-07-27T06:03:22.805756+00:00RAG：基准测试比代码更重要。一个“正确”改进反而降低性能的机器人故事 https://ai-digest-1.com/zh/news/softbank-40-billion-credit-agreement-openai-892AI Digest DNB66zh2026-07-27T06:03:15.389752+00:00软银签署400亿美元信贷协议，联手21家新银行投资OpenAI https://ai-digest-1.com/zh/news/ai-adoption-alone-not-enough-a3-framework-business-change-891AI Digest DNB66zh2026-07-27T06:03:05.846633+00:00仅采用AI不够：实现真正业务变革的A³框架 https://ai-digest-1.com/zh/news/llm-inference-from-kv-cache-to-production-deployment-890AI Digest DNB66zh2026-07-27T06:02:51.218319+00:00LLM推理：从KV缓存到生产部署 https://ai-digest-1.com/zh/news/prx-part-4-data-strategy-diverse-pre-training-corpus-888AI Digest DNB66zh2026-07-27T05:03:52.505601+00:00PRX 第 4 部分：我们的数据策略——构建多样化的预训练语料库 https://ai-digest-1.com/zh/news/nemotron-labs-open-models-ai-trustworthy-controllable-customizable-887AI Digest DNB66zh2026-07-27T05:03:31.659303+00:00Nemotron 实验室：开放模型如何赋予企业和国家可信、可控、可定制的人工智能 https://ai-digest-1.com/zh/news/grok-4-5-for-programming-76-percent-cheaper-than-opus-878AI Digest DNB66zh2026-07-27T05:01:46.756809+00:00编程模型 Grok 4.5：比 Opus 便宜 76% https://ai-digest-1.com/zh/news/nvidia-openai-250-billion-data-center-project-financing-876AI Digest DNB66zh2026-07-27T05:01:30.748680+00:00WSJ：英伟达与OpenAI洽谈2500亿美元数据中心项目融资 https://ai-digest-1.com/zh/news/major-ai-companies-develop-their-own-chips-859AI Digest DNB66zh2026-07-27T04:04:01.696277+00:00主要AI公司大规模自研芯片 https://ai-digest-1.com/zh/news/token-factory-speeds-up-kimi-moment-drives-expansion-of-chinese-computing-power-857AI Digest DNB66zh2026-07-27T04:03:40.596900+00:00"Token 工厂"提速："Kimi 时刻"推动国产算力扩张 https://ai-digest-1.com/zh/news/openai-says-model-carried-out-autonomous-hack-856AI Digest DNB66zh2026-07-27T04:03:29.530692+00:00OpenAI 称其模型实施了自主黑客攻击 https://ai-digest-1.com/zh/news/deepseek-develops-own-chips-blow-to-nvidia-jensen-huang-855AI Digest DNB66zh2026-07-27T04:03:20.909698+00:00DeepSeek自主研发芯片——对英伟达和黄仁勋的又一打击 https://ai-digest-1.com/zh/news/abbel-training-llms-update-beliefs-long-term-interaction-853AI Digest DNB66zh2026-07-27T04:02:58.922870+00:00ABBEL：训练大语言模型更新信念以实现高效长期交互 https://ai-digest-1.com/zh/news/kunlun-wanwei-world-model-paths-852AI Digest DNB66zh2026-07-27T04:02:43.535424+00:00昆仑万维押注两条世界模型路径：实时交互内容与机器人预仿真 https://ai-digest-1.com/zh/news/why-real-ai-agent-deployment-requires-a-gpu-native-cognitive-database-851AI Digest DNB66zh2026-07-27T03:03:51.468287+00:00真正的AI智能体部署为何需要GPU原生认知数据库 https://ai-digest-1.com/zh/news/russian-president-signs-ai-regulation-law-848AI Digest DNB66zh2026-07-27T03:03:26.060737+00:00俄罗斯总统签署人工智能监管法 https://ai-digest-1.com/zh/news/tencent-hunyuan-merges-multimodal-and-llm-departments-yao-shunyu-takes-control-847AI Digest DNB66zh2026-07-27T03:03:19.914355+00:00腾讯混元合并多模态与大模型部门，姚顺宇统一指挥，加速智能体战略 https://ai-digest-1.com/zh/news/ant-bailing-releases-hybrid-reasoning-model-ling-3-0-flash-846AI Digest DNB66zh2026-07-27T03:02:59.765715+00:00蚂蚁百灵发布全新混合推理模型Ling-3.0-Flash https://ai-digest-1.com/zh/news/amd-unveils-helios-2nm-gpu-mi455x-843AI Digest DNB66zh2026-07-27T03:02:26.400724+00:00AMD发布Helios，搭载全球首款2纳米GPU MI455X：OpenAI、Meta和微软等成为客户 https://ai-digest-1.com/zh/news/gct-new-ai-method-brain-prediction-black-box-testable-linguistic-theories-842AI Digest DNB66zh2026-07-27T02:05:57.905842+00:00GCT：新AI方法将大脑预测黑箱转化为可检验的语言学理论 https://ai-digest-1.com/zh/news/google-accelerates-gemini-nano-models-on-pixel-with-frozen-multi-token-predictio-841AI Digest DNB66zh2026-07-27T02:05:49.733125+00:00Google 通过冻结多令牌预测加速 Pixel 上的 Gemini Nano 模型 https://ai-digest-1.com/zh/news/local-ai-coding-agent-qwen3-6-vs-subscriptions-codex-claude-code-840AI Digest DNB66zh2026-07-27T02:05:38.991019+00:00本地AI编程助手：Qwen3.6搭配Qwen-Code vs Codex和Claude Code订阅 https://ai-digest-1.com/zh/news/self-learning-nvidia-robots-tencent-gpu-cluster-elegiac-essay-839AI Digest DNB66zh2026-07-27T02:05:28.900972+00:00自学习NVIDIA机器人、腾讯万GPU集群，以及关于人类未来的挽歌式文章 https://ai-digest-1.com/zh/news/memora-harmonic-memory-abstraction-specificity-838AI Digest DNB66zh2026-07-27T02:05:20.501633+00:00Memora：平衡抽象与细节的谐波记忆表征 https://ai-digest-1.com/zh/news/google-tabfm-foundation-model-tabular-data-837AI Digest DNB66zh2026-07-27T02:05:09.462931+00:00谷歌推出 TabFM：用于表格数据的零样本学习基础模型 https://ai-digest-1.com/zh/news/skillopt-ai-agent-skills-trainable-parameters-835AI Digest DNB66zh2026-07-27T02:04:49.547475+00:00SkillOpt：将AI智能体技能转化为可训练参数 https://ai-digest-1.com/zh/news/mistral-ai-leanstral-1-5-830AI Digest DNB66zh2026-07-27T02:04:03.592689+00:00Mistral AI 发布 Leanstral 1.5：免费的形式验证模型，创纪录结果 https://ai-digest-1.com/zh/news/run-ai-workloads-on-any-cloud-with-hugging-face-storage-zero-egress-via-skypilot-826AI Digest DNB66zh2026-07-27T02:03:31.120945+00:00通过 Hugging Face Storage 在任何云上运行 AI 工作负载：SkyPilot 实现零出站费用 https://ai-digest-1.com/zh/news/samsung-lee-jae-yong-met-sam-altman-ai-infrastructure-cooperation-821AI Digest DNB66zh2026-07-27T02:02:31.091494+00:00三星李在镕会见山姆·奥尔特曼，讨论扩大AI基础设施合作 https://ai-digest-1.com/zh/news/ai-model-benchmark-1c-development-updated-ranking-v2-820AI Digest DNB66zh2026-07-27T02:02:19.169932+00:001C开发AI模型基准测试：更新排名v2 https://ai-digest-1.com/zh/news/nvidia-guarantor-softbank-500-billion-data-center-ohio-818AI Digest DNB66zh2026-07-27T02:02:05.556409+00:00英伟达将为软银在俄亥俄州5000亿美元的数据中心项目提供担保 https://ai-digest-1.com/zh/news/nvidia-uses-vera-cpu-to-accelerate-new-chip-design-817AI Digest DNB66zh2026-07-27T02:02:00.168623+00:00Nvidia将Vera CPU用于加速新一代CPU和GPU的设计 https://ai-digest-1.com/zh/news/fable-gpu-cores-ai-automation-analog-computing-import-ai-464-815AI Digest DNB66zh2026-07-27T01:07:01.588712+00:00Fable编写GPU核心、AI自动化与模拟计算：Import AI 464文摘 https://ai-digest-1.com/zh/news/lerobot-v0-6-0-world-model-policies-vla-benchmarks-cli-813AI Digest DNB66zh2026-07-27T01:06:35.046336+00:00LeRobot v0.6.0：基于世界模型的策略、新VLA、基准测试与评估CLI https://ai-digest-1.com/zh/news/intelligence-becomes-free-whats-next-data-systems-for-agents-of-agents-and-by-ag-812AI Digest DNB66zh2026-07-27T01:06:27.036604+00:00智能变得免费——接下来怎么办？智能体数据系统，智能体的数据系统，以及由智能体构建的数据系统 https://ai-digest-1.com/zh/news/hugging-face-microsoft-foundry-open-models-managed-gpus-811AI Digest DNB66zh2026-07-27T01:06:14.004727+00:00Hugging Face 与 Microsoft Foundry：在托管 GPU 上运行开放模型 https://ai-digest-1.com/zh/news/new-tool-reverses-loss-of-programming-skills-through-vibe-coding-809AI Digest DNB66zh2026-07-27T01:05:59.076157+00:00新工具承诺通过“氛围编程”逆转编程技能丧失 https://ai-digest-1.com/zh/news/hugging-face-transformers-backend-vllm-native-performance-807AI Digest DNB66zh2026-07-27T01:05:40.206442+00:00Hugging Face Transformers 后端集成至 vLLM，实现原生性能 https://ai-digest-1.com/zh/news/ai-overthinking-security-vulnerability-806AI Digest DNB66zh2026-07-27T01:05:32.768609+00:00AI模型易‘过度思考’引发安全漏洞 https://ai-digest-1.com/zh/news/mistral-ai-robostral-navigate-robot-navigation-rgb-camera-805AI Digest DNB66zh2026-07-27T01:05:25.960338+00:00Mistral AI推出Robostral Navigate：仅用单个RGB摄像头实现机器人导航的模型 https://ai-digest-1.com/zh/news/finam-ai-lab-updates-finesse-bench-for-llms-803AI Digest DNB66zh2026-07-27T01:05:05.430632+00:00Finam AI实验室更新金融基准FINESSE-Bench用于大语言模型 https://ai-digest-1.com/zh/news/gpt-5-6-becomes-main-model-microsoft-365-copilot-801AI Digest DNB66zh2026-07-27T01:04:52.435374+00:00GPT-5.6 成为 Microsoft 365 Copilot 的主要模型 https://ai-digest-1.com/zh/news/from-three-nos-to-billion-dollar-ipo-liang-wenfeng-deepseek-three-year-battle-799AI Digest DNB66zh2026-07-27T01:04:30.497765+00:00从“三不”原则到百亿IPO传闻：梁文锋与DeepSeek的三年鏖战 https://ai-digest-1.com/zh/news/openai-model-attacks-hugging-face-zhipu-ai-fixes-vulnerability-798AI Digest DNB66zh2026-07-27T01:04:22.549497+00:00OpenAI模型测试期间攻击Hugging Face社区；智谱AI开源模型修复漏洞 https://ai-digest-1.com/zh/news/savings-on-llms-chinese-model-writes-chinese-793AI Digest DNB66zh2026-07-27T01:03:37.821797+00:00LLM成本节省：中国模型竟自动输出汉字 https://ai-digest-1.com/zh/news/openai-agent-hack-discovered-a-week-after-attack-790AI Digest DNB66zh2026-07-27T01:03:02.456786+00:00消息来源：OpenAI 在攻击发生仅一周后发现其智能体被黑客入侵 https://ai-digest-1.com/zh/news/anthropic-rethinks-context-engineering-for-claude-5-788AI Digest DNB66zh2026-07-27T01:02:45.047837+00:00Anthropic 为 Claude 5 重新思考上下文工程：简化指令效果更好 https://ai-digest-1.com/zh/news/google-sensorfm-foundation-model-wearables-trillion-minutes-786AI Digest DNB66zh2026-07-27T00:04:52.552423+00:00SensorFM：基于万亿分钟数据训练的可穿戴设备基础模型 https://ai-digest-1.com/zh/news/mistral-studio-unified-prompt-skill-management-785AI Digest DNB66zh2026-07-27T00:04:44.616325+00:00Mistral Studio 推出统一提示词与技能管理 https://ai-digest-1.com/zh/news/large-tabular-models-outperform-llms-on-structured-data-784AI Digest DNB66zh2026-07-27T00:04:36.739573+00:00大型表格模型在结构化数据上超越大语言模型 https://ai-digest-1.com/zh/news/microsoft-released-aurora-1-5-open-foundation-model-weather-earth-systems-783AI Digest DNB66zh2026-07-27T00:04:29.631824+00:00微软发布Aurora 1.5：面向天气和地球系统的开源基础模型 https://ai-digest-1.com/zh/news/reference-architecture-enterprise-ai-harness-kubernetes-781AI Digest DNB66zh2026-07-27T00:04:07.512330+00:00企业级AI Harness在Kubernetes上的参考架构：生产环境中安全部署代理的四层模型 https://ai-digest-1.com/zh/news/mamba-architecture-transformers-hybrid-models-780AI Digest DNB66zh2026-07-27T00:03:59.821845+00:00Mamba：旨在取代Transformer的架构 https://ai-digest-1.com/zh/news/quantcode-bench-evaluating-llms-algorithmic-trading-strategies-779AI Digest DNB66zh2026-07-27T00:03:50.009257+00:00QuantCode-Bench：评估大语言模型生成可执行算法交易策略能力的新基准 https://ai-digest-1.com/zh/news/x-square-robot-general-purpose-robots-stack-778AI Digest DNB66zh2026-07-27T00:03:41.443255+00:00X Square Robot 构建通用机器人基础栈 https://ai-digest-1.com/zh/news/thinking-machines-releases-inkling-1-trillion-parameters-776AI Digest DNB66zh2026-07-27T00:03:21.747115+00:00Thinking Machines 发布拥有1万亿参数的开源多模态模型 Inkling https://ai-digest-1.com/zh/news/linus-torvalds-linux-not-anti-ai-useful-tool-775AI Digest DNB66zh2026-07-27T00:03:14.415728+00:00Linus Torvalds：Linux 并非反 AI 项目，而是一个有用的工具 https://ai-digest-1.com/zh/news/zhipu-stock-surges-hugging-face-glm-5-2-deployment-773AI Digest DNB66zh2026-07-27T00:02:54.536864+00:00智谱股价暴涨超30%：Hugging Face 披露GLM-5.2部署实例，AI竞争进入“模型+基础设施”时代 https://ai-digest-1.com/zh/news/liang-wenfeng-deepseek-does-not-aim-to-become-the-next-bytedance-or-tencent-low-772AI Digest DNB66zh2026-07-27T00:02:47.535989+00:00梁文锋：DeepSeek无意成为下一个字节跳动或腾讯，低价策略很简单 https://ai-digest-1.com/zh/news/kimi-k3-attracts-attention-ubs-structural-price-advantage-chinese-models-768AI Digest DNB66zh2026-07-27T00:02:14.250406+00:00Kimi K3 再次引发关注：瑞银认为中国模型的结构性价格优势将持续 https://ai-digest-1.com/zh/news/nvidia-negotiating-openai-data-center-guarantees-767AI Digest DNB66zh2026-07-27T00:01:54.132441+00:00媒体：英伟达正就OpenAI数据中心2500亿美元担保进行谈判 https://ai-digest-1.com/zh/news/performance-per-watt-key-metric-for-ai-infrastructure-efficiency-765AI Digest DNB66zh2026-07-26T23:05:08.335081+00:00能效比是AI基础设施效率的关键指标 https://ai-digest-1.com/zh/news/how-i-turned-ai-to-the-dark-side-systemic-vulnerabilities-in-leading-llms-764AI Digest DNB66zh2026-07-26T23:04:57.738881+00:00我是如何将AI引向黑暗面的：主流大型语言模型中的系统性漏洞 https://ai-digest-1.com/zh/news/ai-folds-dna-into-miniature-masterpieces-761AI Digest DNB66zh2026-07-26T23:04:30.582286+00:00AI 将DNA折叠成微型杰作 https://ai-digest-1.com/zh/news/domain-routed-agents-vs-naive-rag-comparison-759AI Digest DNB66zh2026-07-26T23:04:11.577286+00:00领域路由智能体与朴素RAG在桌面角色扮演专家系统任务中的比较 https://ai-digest-1.com/zh/news/model-routing-simple-until-reality-758AI Digest DNB66zh2026-07-26T23:04:02.846216+00:00模型路由：看似简单，直到遇到现实 https://ai-digest-1.com/zh/news/building-shippy-lessons-reliable-ai-agents-operational-tasks-757AI Digest DNB66zh2026-07-26T23:03:52.254266+00:00构建Shippy教会我们的：为操作任务构建可靠AI代理的经验 https://ai-digest-1.com/zh/news/nvidia-jetson-thor-t3000-t2000-robotics-edge-ai-755AI Digest DNB66zh2026-07-26T23:03:35.572551+00:00NVIDIA 推出面向大众市场机器人与边缘AI的Jetson Thor T3000和T2000 https://ai-digest-1.com/zh/news/hugging-face-2026-breach-ai-agent-details-753AI Digest DNB66zh2026-07-26T23:03:20.968649+00:00Hugging Face 披露2026年7月涉及AI代理的安全事件详情 https://ai-digest-1.com/zh/news/inkling-open-model-from-thinking-machines-lab-752AI Digest DNB66zh2026-07-26T23:03:15.081736+00:00Inkling：Thinking Machines Lab 发布的开源模型 https://ai-digest-1.com/zh/news/kimi-k3-2-8-trillion-parameters-pelican-benchmark-750AI Digest DNB66zh2026-07-26T23:02:56.236280+00:00Kimi K3：2.8万亿参数以及我们仍能从鹈鹕基准测试中学到什么 https://ai-digest-1.com/zh/news/firefox-in-webassembly-running-browser-inside-browser-749AI Digest DNB66zh2026-07-26T23:02:48.773675+00:00WebAssembly 中的 Firefox：在浏览器中运行浏览器 https://ai-digest-1.com/zh/news/ai-attack-chinese-large-model-glm-4-rescue-748AI Digest DNB66zh2026-07-26T23:02:31.915323+00:00AI失控攻击企业系统：中国大模型挺身解围——6月中旬上线，权威榜单中达到开源模型最佳水平 https://ai-digest-1.com/zh/news/cognition-acquires-poke-ai-personality-competitive-advantage-746AI Digest DNB66zh2026-07-26T23:02:10.946651+00:00Cognition为何收购Poke：AI人格成为竞争优势 https://ai-digest-1.com/zh/news/ai-agent-smartphones-nubia-stepfun-honor-competition-745AI Digest DNB66zh2026-07-26T23:01:55.764396+00:00AI智能体手机：努比亚、阶跃星辰与荣耀竞逐——模型、硬件与应用的暗战 https://ai-digest-1.com/zh/news/deepmind-isomorphic-labs-bioresilience-strategy-743AI Digest DNB66zh2026-07-26T22:05:54.704744+00:00Google DeepMind 与 Isomorphic Labs 联合发布生物弹性战略 https://ai-digest-1.com/zh/news/chartnet-mit-ai-understand-charts-million-scale-dataset-741AI Digest DNB66zh2026-07-26T22:05:35.590688+00:00MIT 的 ChartNet：用百万级数据集教 AI 理解图表 https://ai-digest-1.com/zh/news/eu-forces-google-to-share-search-data-and-open-android-to-third-party-ai-assista-738AI Digest DNB66zh2026-07-26T22:05:04.919691+00:00欧盟强制谷歌共享搜索数据并开放安卓系统给第三方AI助手 https://ai-digest-1.com/zh/news/grok-4-3-from-xai-now-available-on-amazon-bedrock-737AI Digest DNB66zh2026-07-26T22:04:57.669013+00:00xAI 的 Grok 4.3 现已在 Amazon Bedrock 上可用 https://ai-digest-1.com/zh/news/huawei-alibaba-baidu-supernodes-1024-gpus-10-trillion-parameters-734AI Digest DNB66zh2026-07-26T22:04:26.596822+00:00华为、阿里和百度联合推出超级节点：算力高达1024块GPU，支持万亿参数模型推理 https://ai-digest-1.com/zh/news/deepseek-ipo-kimi-trillion-parameter-model-chinese-ai-new-stage-733AI Digest DNB66zh2026-07-26T22:04:18.615693+00:00DeepSeek 筹备IPO，Kimi 发布万亿参数模型：中国AI进入新阶段 https://ai-digest-1.com/zh/news/moonshot-ai-kimi-k3-tops-global-ranking-bloomberg-says-us-china-ai-gap-narrows-731AI Digest DNB66zh2026-07-26T22:04:07.458998+00:00Moonshot AI 的 Kimi K3 登顶全球排名：彭博称中美人工智能差距缩小 https://ai-digest-1.com/zh/news/has-deepseeks-basic-logic-changed-727AI Digest DNB66zh2026-07-26T22:03:35.155897+00:00DeepSeek的基本逻辑变了吗？ https://ai-digest-1.com/zh/news/liang-wenfeng-deepseek-openai-interview-726AI Digest DNB66zh2026-07-26T22:03:30.301546+00:00梁文锋三小时访谈：为何DeepSeek在OpenAI无法成功的路径上取得成功？ https://ai-digest-1.com/zh/news/from-one-factory-to-two-story-harness-dev-pipeline-grew-in-under-a-year-723AI Digest DNB66zh2026-07-26T22:02:58.239994+00:00从单一工厂到双层引擎：一条开发流水线如何在不到一年内成长 https://ai-digest-1.com/zh/news/grok-4-5-benchmarks-founder-payments-721AI Digest DNB66zh2026-07-26T22:02:45.829528+00:00Grok 4.5：基准测试不敌对手，却赢得创始人支票 https://ai-digest-1.com/zh/news/ai-model-availability-26-july-2026-720AI Digest DNB66zh2026-07-26T21:14:04.352707+00:002026年7月26日AI模型可用性 https://ai-digest-1.com/zh/news/sam-altman-open-source-gpt-3-model-letter-2022-715AI Digest DNB66zh2026-07-26T21:03:41.016704+00:00山姆·奥尔特曼曾想在2022年发布GPT-3级别的开源模型以阻碍竞争对手 https://ai-digest-1.com/zh/news/rumor-deepseek-v4-full-version-launch-today-can-it-surpass-kimi-k3-712AI Digest DNB66zh2026-07-26T21:03:25.840781+00:00传闻：DeepSeek V4 "完整版"今日发布，能否超越Kimi K3？ https://ai-digest-1.com/zh/news/moonshot-ai-suspends-kimi-subscriptions-707AI Digest DNB66zh2026-07-26T21:02:52.543232+00:00Moonshot AI 暂停Kimi订阅：模型效率提升，但算力缺口扩大 https://ai-digest-1.com/zh/news/vivix-a1-real-time-virtual-character-communication-704AI Digest DNB66zh2026-07-26T21:02:13.255928+00:00Vivix 发布 A1：首个用于交互式虚拟角色通信的实时模型 https://ai-digest-1.com/zh/news/xiaomi-registers-10-ai-models-and-integrates-ai-with-own-chips-and-system-703AI Digest DNB66zh2026-07-26T21:02:06.115177+00:00小米注册10个AI模型，将AI与其自有芯片和系统集成 https://ai-digest-1.com/zh/news/apple-intelligence-china-iphone-digital-sovereignty-decline-698AI Digest DNB66zh2026-07-26T20:04:59.606671+00:00苹果智能终入中国，但iPhone的“数字主权”开始削弱 https://ai-digest-1.com/zh/news/doubao-roewe-ai-native-design-car-696AI Digest DNB66zh2026-07-26T20:04:32.336382+00:00豆包与荣威联手发布全球首款AI原生设计汽车 https://ai-digest-1.com/zh/news/businesses-may-be-banned-from-using-chatgpt-and-deepseek-692AI Digest DNB66zh2026-07-26T20:03:54.330797+00:00企业可能被禁止使用ChatGPT和DeepSeek：新法律后如何处理外国AI服务 https://ai-digest-1.com/zh/news/ai-model-price-war-second-phase-inference-cost-reduction-691AI Digest DNB66zh2026-07-26T20:03:47.809455+00:00AI模型价格战进入第二阶段：推理成本降低 https://ai-digest-1.com/zh/news/ai-and-law-legal-departments-690AI Digest DNB66zh2026-07-26T20:03:42.069195+00:00人工智能与法律：法律部门需要了解什么 https://ai-digest-1.com/zh/news/trump-administration-rule-limits-public-input-data-center-permits-689AI Digest DNB66zh2026-07-26T20:03:31.940228+00:00特朗普政府考虑限制公众对数据中心许可的参与 https://ai-digest-1.com/zh/news/openai-gpt5-6-sol-terra-luna-on-amazon-bedrock-688AI Digest DNB66zh2026-07-26T20:03:18.986887+00:00OpenAI GPT-5.6 Sol、Terra 和 Luna 模型现已在 Amazon Bedrock 上可用 https://ai-digest-1.com/zh/news/katy-isd-limits-ai-in-elementary-schools-687AI Digest DNB66zh2026-07-26T20:03:10.558644+00:00凯蒂独立学区限制小学使用人工智能 https://ai-digest-1.com/zh/news/openai-prepares-voice-interface-revolution-686AI Digest DNB66zh2026-07-26T20:02:52.161444+00:00OpenAI 准备掀起语音界面革命：新设备和ChatGPT语音模式 https://ai-digest-1.com/zh/news/self-hosting-vs-api-cost-of-solved-task-open-models-684AI Digest DNB66zh2026-07-26T20:02:29.948415+00:00自托管与API：在开源模型时代如何计算解决任务的实际成本 https://ai-digest-1.com/zh/news/eliza-not-just-regex-the-architectural-depth-of-the-first-chatbot-683AI Digest DNB66zh2026-07-26T20:02:19.790277+00:00ELIZA：不仅仅是正则——首个聊天机器人的架构深度 https://ai-digest-1.com/zh/news/panic-around-chinese-ai-deja-vu-682AI Digest DNB66zh2026-07-26T20:02:07.048269+00:00围绕中国AI的恐慌：似曾相识？ https://ai-digest-1.com/zh/news/meta-free-ai-model-distribution-weakens-competitors-680AI Digest DNB66zh2026-07-26T19:03:24.437416+00:00Meta免费分发AI模型意在削弱竞争对手 https://ai-digest-1.com/zh/news/large-language-models-interpreting-medical-examination-results-nature-679AI Digest DNB66zh2026-07-26T19:03:17.976463+00:00大型语言模型用于解读医学检查结果 ——《自然》 https://ai-digest-1.com/zh/news/moonshot-ai-final-funding-round-before-ipo-50-billion-valuation-677AI Digest DNB66zh2026-07-26T19:03:06.869808+00:00中国AI开发商月之暗面在IPO前计划最后一轮融资，估值500亿美元 https://ai-digest-1.com/zh/news/waic-aigpt-deepseek-fashion-ai-676AI Digest DNB66zh2026-07-26T19:03:01.628096+00:00WAIC专访：AIGPT、DeepSeek虽强，但时尚品牌更需要懂行业的AI https://ai-digest-1.com/zh/news/glm-5-explodes-market-zhipu-shares-surge-domestic-large-models-accelerate-674AI Digest DNB66zh2026-07-26T19:02:51.599204+00:00GLM-5引爆市场！智谱股价飙升28%，国产大模型春节前加速发布 https://ai-digest-1.com/zh/news/openai-partners-with-us-department-of-energy-national-labs-672AI Digest DNB66zh2026-07-26T19:02:35.831280+00:00OpenAI宣布与美国能源部及国家实验室合作，加速科学发现 https://ai-digest-1.com/zh/news/battle-of-giants-bytedance-alibaba-jieyue-reshape-audio-economy-671AI Digest DNB66zh2026-07-26T19:02:25.178583+00:00巨人战争：字节跳动、阿里巴巴与捷跃重塑120亿音频经济 https://ai-digest-1.com/zh/news/deepseek-busy-before-ipo-liang-wenfeng-tops-richest-ai-founders-list-669AI Digest DNB66zh2026-07-26T19:02:08.025290+00:00DeepSeek IPO前繁忙：梁文锋登顶AI公司创始人富豪榜 https://ai-digest-1.com/zh/news/baidu-yijing-digital-human-waic-667AI Digest DNB66zh2026-07-26T19:01:46.756178+00:00数字人获得“活”的感觉：百度文心一言的WAIC时刻 https://ai-digest-1.com/zh/news/agroeco-timiryazev-academy-test-russian-ai-early-detection-lameness-cows-666AI Digest DNB66zh2026-07-26T19:01:38.347523+00:00AgroEco与季米里亚泽夫农学院测试俄罗斯AI系统早期检测奶牛跛行 https://ai-digest-1.com/zh/news/meta-ai-muse-spark-1-1-update-665AI Digest DNB66zh2026-07-26T19:01:21.904948+00:00Meta AI 将音乐生成器 Muse Spark 更新至 1.1 版本 https://ai-digest-1.com/zh/news/kimi-k3-china-open-source-world-class-model-664AI Digest DNB66zh2026-07-26T18:04:47.492676+00:00Kimi K3：中国开源世界级模型——它为何强大？ https://ai-digest-1.com/zh/news/spacexai-deployable-apps-for-grok-build-663AI Digest DNB66zh2026-07-26T18:04:41.727644+00:00SpaceXAI 为 Grok Build 开发可部署应用 https://ai-digest-1.com/zh/news/fine-tuning-rugpt-3-xl-1-3b-to-modern-llm-level-comparison-with-gemma3-1b-659AI Digest DNB66zh2026-07-26T18:04:17.511293+00:00将 ruGPT-3 XL 1.3B 微调至现代大语言模型水平，并与 Gemma3 1B 对比 https://ai-digest-1.com/zh/news/shanghai-ai-agreement-details-published-656AI Digest DNB66zh2026-07-26T18:04:01.853729+00:00上海人工智能协议细节公布 https://ai-digest-1.com/zh/news/report-ai-attacks-ai-first-automated-cyberattack-by-openai-repelled-by-chinese-m-655AI Digest DNB66zh2026-07-26T18:03:45.597154+00:00报告：AI攻击AI——OpenAI发起世界首次自动化网络攻击，被中国智谱GLM-5模型击退 https://ai-digest-1.com/zh/news/helper-russian-corpus-psychological-text-analysis-qwen-fine-tuning-654AI Digest DNB66zh2026-07-26T18:03:31.453641+00:00HELPER：用于心理文本分析和本地情绪检测的俄语语料库及Qwen微调 https://ai-digest-1.com/zh/news/waico-new-international-organization-for-ai-control-653AI Digest DNB66zh2026-07-26T18:03:25.767694+00:00WAICO：新成立的人工智能控制国际组织 https://ai-digest-1.com/zh/news/ai-oled-blue-color-molecules-652AI Digest DNB66zh2026-07-26T18:03:20.216602+00:00AI助力寻找OLED屏幕理想蓝色分子 https://ai-digest-1.com/zh/news/liang-wenfeng-names-8-ai-companies-worth-watching-651AI Digest DNB66zh2026-07-26T18:03:12.965274+00:00梁文峰点名8家值得关注的AI公司 https://ai-digest-1.com/zh/news/snake-robots-searched-for-survivors-after-venezuela-earthquake-649AI Digest DNB66zh2026-07-26T18:02:58.625709+00:00蛇形机器人在委内瑞拉地震后搜寻废墟下的幸存者 https://ai-digest-1.com/zh/news/geforce-rtx-3050-doubles-frame-rate-of-rtx-3090-via-lossless-scaling-648AI Digest DNB66zh2026-07-26T18:02:52.814887+00:00GeForce RTX 3050 通过无损缩放技术让 GeForce RTX 3090 帧率翻倍 https://ai-digest-1.com/zh/news/arkansas-family-sues-xai-over-grok-deepfake-child-abuse-647AI Digest DNB66zh2026-07-26T18:02:43.125070+00:00阿肯色州家庭起诉xAI，指控其模型Grok被用于制作深度伪造儿童虐待内容 https://ai-digest-1.com/zh/news/ai-vs-ai-agent-runtime-security-646AI Digest DNB66zh2026-07-26T18:02:28.972254+00:00AI对抗AI：每次使用都会更强的智能体运行时安全方案 https://ai-digest-1.com/zh/news/librarians-hold-how-to-avoid-ai-workshops-645AI Digest DNB66zh2026-07-26T18:02:21.389435+00:00图书馆员举办爆款讲习班：教厌倦大科技的人‘如何避开人工智能’ https://ai-digest-1.com/zh/news/black-forest-labs-flux-3-multimodal-model-644AI Digest DNB66zh2026-07-26T18:02:08.882812+00:00Black Forest Labs 发布 FLUX 3：面向图像、视频、音频和机器人动作预测的多模态模型