{"id":18549,"date":"2026-08-28T08:58:52","date_gmt":"2026-08-27T22:58:52","guid":{"rendered":"https:\/\/geoffreychen.com\/ai-technology-observation-2026-08-28-08\/"},"modified":"2026-08-28T08:58:52","modified_gmt":"2026-08-27T22:58:52","slug":"ai-technology-observation-2026-08-28-08","status":"publish","type":"post","link":"https:\/\/geoffreychen.com\/zh\/ai-technology-observation-2026-08-28-08\/","title":{"rendered":"AI\u6280\u672f\u89c2\u5bdf\uff5c2026\u5e748\u670828\u65e5 08\u65f6"},"content":{"rendered":"<\/p>\n<p>\u672c\u6b21\u89c2\u5bdf\u4e3b\u8981\u8986\u76d62026\u5e748\u670827\u65e508\u65f647\u5206\u81f38\u670828\u65e508\u65f647\u5206\uff08Australia\/Sydney\uff09\uff0c\u6536\u5f555\u9879\u5df2\u7ecf\u516c\u5f00\u5e76\u53ef\u6838\u9a8c\u7684\u6280\u672f\u53d8\u5316\u3002<\/p>\n<p>Anthropic\u5f00\u653eModel Hardware Standard\u7814\u7a76\u9884\u89c8<\/p>\n<p>Anthropic\u4e8e8\u670827\u65e5\u5411\u9996\u6279\u79d1\u7814\u5b9e\u9a8c\u5ba4\u548c\u5148\u8fdb\u5236\u9020\u673a\u6784\u5f00\u653eModel Hardware Standard\uff08MHS\uff09\u7814\u7a76\u9884\u89c8\u3002MHS\u7528\u7edf\u4e00\u9a71\u52a8\u7a0b\u5e8f\u3001\u8bfb\u5199\u539f\u8bed\u548c\u8bbe\u5907\u63cf\u8ff0\u6587\u4ef6\u8fde\u63a5\u5177\u6709\u53ef\u7f16\u7a0b\u63a5\u53e3\u7684\u4eea\u5668\uff0c\u4ee3\u7406\u53ef\u7ecfMCP\u3001\u547d\u4ee4\u884c\u6216API\u63a7\u5236\u591a\u53f0\u8bbe\u5907\u3002\u76f8\u8f83\u6b64\u524d\u9010\u53f0\u7f16\u5199\u96c6\u6210\u4ee3\u7801\uff0c\u65b0\u589e\u4e8b\u5b9e\u662f\u4e00\u4e2a\u6a21\u578b\u65e0\u5173\u7684\u786c\u4ef6\u63a5\u53e3\u5df2\u5728\u663e\u5fae\u955c\u3001\u79fb\u6db2\u8bbe\u5907\u3001\u673a\u68b0\u81c2\u548c\u91cf\u5b50\u6fc0\u5149\u7cfb\u7edf\u4e2d\u8bd5\u7528\u3002QuEra\u786e\u8ba4\uff0cClaude\u901a\u8fc7MHS\u5728\u7ea670\u4e07\u7f8e\u5143\u7684\u6d4b\u8bd5\u8bbe\u5907\u4e0a\u5f00\u53d1\u5e76\u9a8c\u8bc1\u4e86\u6fc0\u5149\u81ea\u52a8\u91cd\u9501\u63a7\u5236\u5668\uff1b\u6295\u5165\u8fd0\u884c\u7684\u662f\u53ef\u68c0\u67e5\u7684\u786e\u5b9a\u6027\u7a0b\u5e8f\uff0cAI\u4e0d\u5728\u81ea\u52a8\u91cd\u9501\u8fd0\u884c\u73af\u8def\u4e2d\uff0c\u8c03\u8c10\u6d41\u7a0b\u4ecd\u4fdd\u7559AI\u3002\u8be5\u6807\u51c6\u76ee\u524d\u4ec5\u9650\u7814\u7a76\u9884\u89c8\uff0c\u5c1a\u672a\u5f00\u6e90\uff0c\u7269\u7406\u63a8\u7406\u9650\u5236\u3001\u5b89\u5168\u8bc4\u6d4b\u548c\u66f4\u591a\u8bbe\u5907\u9a71\u52a8\u4ecd\u672a\u5b8c\u6210\u3002<\/p>\n<p>Google\u53d1\u5e03Gemini Omni 1.1 Flash<\/p>\n<p>Google\u4e8e8\u670827\u65e5\u53d1\u5e03Gemini Omni 1.1 Flash\uff0c\u5e76\u901a\u8fc7Google AI Studio\u7684Gemini API\u548cGemini Enterprise Agent Platform\u9010\u6b65\u5f00\u653e\u3002\u65b0\u7248\u672c\u628a\u89c6\u9891\u5ef6\u5c55\u6240\u53c2\u8003\u7684\u65e2\u6709\u753b\u9762\u4ece\u6700\u540e1\u79d2\u589e\u52a0\u5230\u6700\u591a10\u79d2\uff0c\u53ef\u630910\u79d2\u9012\u589e\u5ef6\u5c55\u5230\u7d2f\u8ba140\u79d2\uff0c\u5e76\u52a0\u5165\u9996\u5c3e\u5173\u952e\u5e27\u63d2\u503c\u3001360p\u8349\u7a3f\u548c\u6700\u9ad84K\u653e\u5927\u3002360p\u751f\u6210\u201c\u6700\u9ad8\u5feb60%\u4e14\u6210\u672c\u4e3a720p\u4e09\u5206\u4e4b\u4e00\u201d\u7b49\u6570\u636e\u6765\u81eaGoogle\uff1b\u76ee\u524d\u6ca1\u6709\u72ec\u7acb\u8d28\u91cf\u8bc4\u6d4b\uff0c\u5b9e\u9645\u4e00\u81f4\u6027\u548c\u590d\u6742\u955c\u5934\u63a7\u5236\u80fd\u529b\u4ecd\u4ee5\u5f00\u53d1\u8005\u62a5\u544a\u4e3a\u51c6\u3002<\/p>\n<p>Gemini 3.5 Transcribe\u8fdb\u5165API\u516c\u6d4b<\/p>\n<p>Google\u4e8e8\u670826\u65e5\u53d1\u5e03Gemini 3.5 Transcribe\u548c\u5b9e\u65f6\u7248\u672cgemini-3.5-transcribe-live\u3002\u4e24\u8005\u5206\u522b\u901a\u8fc7Interactions API\u548cLive API\u63d0\u4f9b\uff0c\u652f\u630185\u79cd\u4ee5\u4e0a\u8bed\u8a00\u3001\u672f\u8bed\u8bcd\u8868\u548c\u81ea\u52a8\u6574\u7406\u6587\u672c\uff1b\u6587\u4ef6\u6a21\u5f0f\u8fd8\u652f\u6301\u6700\u591a\u4e09\u4eba\u7684\u8bf4\u8bdd\u4eba\u533a\u5206\u4e0e\u9010\u8bcd\u65f6\u95f4\u6233\u3002\u5f00\u53d1\u8005\u548c\u4f01\u4e1a\u63a5\u53e3\u5747\u4e3a\u516c\u5f00\u9884\u89c8\uff0cmacOS Gemini\u5e94\u7528\u4ec5\u5f00\u653e\u82f1\u6587\uff0cAndroid Rambler\u9650\u90e8\u5206\u56fd\u5bb6\u548c\u8bed\u8a00\uff0cChrome\u5c1a\u672a\u4e0a\u7ebf\u3002Artificial Analysis\u72ec\u7acb\u6d4b\u8bd5\u8bb0\u5f55\u6587\u4ef6\u6a21\u5f0f2.6%\u7684AA-WER\u3001\u5b9e\u65f6\u6a21\u5f0f4.0%\u7684AA-WER\uff1bGoogle\u6240\u79f0\u76f8\u5bf9Chirp 3\u6700\u7ec8\u8f6c\u5199\u65f6\u95f4\u7f29\u77ed70%\u4ecd\u5c5e\u4e8e\u5f00\u53d1\u8005\u6bd4\u8f83\u3002<\/p>\n<p>NVIDIA Vera CPU\u4ece\u65e9\u671f\u4ea4\u4ed8\u8f6c\u5411\u89c4\u6a21\u51fa\u8d27<\/p>\n<p>NVIDIA\u57288\u670827\u65e5\u66f4\u65b0\u51765\u670818\u65e5\u6750\u6599\uff0c\u79f0\u9996\u6b3e\u81ea\u7814Vera CPU\u5df2\u5f00\u59cb\u89c4\u6a21\u51fa\u8d27\uff0c\u5e76\u786e\u8ba4AWS\u6536\u5230\u9996\u53f0Vera CPU\u670d\u52a1\u5668\u548c\u4e00\u679aVera Rubin GPU\uff1b\u6b64\u524d\u4ea4\u4ed8\u5bf9\u8c61\u5305\u62ecOCI\u3001Anthropic\u3001OpenAI\u548cSpaceXAI\u3002Vera\u542b88\u4e2aOlympus\u6838\u5fc3\u548c1.2TB\/s\u5185\u5b58\u5e26\u5bbd\uff0c\u9762\u5411\u4ee3\u7406\u7f16\u6392\u3001\u5de5\u5177\u8c03\u7528\u3001\u6c99\u7bb1\u548c\u5f3a\u5316\u5b66\u4e60\u7b49CPU\u8d1f\u8f7d\u3002\u6700\u9ad81.8\u500d\u6bcf\u6838\u6027\u80fd\u6765\u81eaNVIDIA\u6d4b\u8bd5\uff0c\u5c1a\u65e0\u516c\u5f00\u72ec\u7acb\u590d\u73b0\u3002AWS\u786e\u8ba4\u5c06\u5f15\u5165Vera\u57fa\u7840\u8bbe\u65bd\uff0c\u4f46\u672a\u7ed9\u51fa\u9762\u5411\u4e91\u5ba2\u6237\u7684\u6b63\u5f0f\u53ef\u7528\u65e5\u671f\uff1b\u5df2\u4ea4\u4ed8\u6d4b\u8bd5\u7cfb\u7edf\u4e0d\u80fd\u7b49\u540c\u4e8eAWS\u670d\u52a1\u5168\u9762\u5f00\u653e\u3002<\/p>\n<p>Pollen Robotics\u5f00\u653eMicroduck\u9884\u8ba2\u4e0e\u5f00\u6e90\u6808<\/p>\n<p>Hugging Face\u65d7\u4e0bPollen Robotics\u4e8e8\u670827\u65e5\u5f00\u653eMicroduck\u9884\u8ba2\uff0c\u4ef7\u683c\u4e3a399\u7f8e\u5143\uff0c\u9996\u6279\u4ea4\u4ed8\u76ee\u6807\u4e3a2026\u5e74\u5723\u8bde\u8282\u524d\uff0c\u521d\u671f\u5e02\u573a\u4e3a\u5317\u7f8e\u548c\u6b27\u6d32\u3002\u8be525\u5398\u7c73\u53cc\u8db3\u673a\u5668\u4eba\u914d\u670915\u4e2a\u7535\u673a\u3001\u6444\u50cf\u5934\u3001LiDAR\u3001\u4e24\u7ec4IMU\u548c\u53ef\u6293\u53d6\u7269\u4f53\u7684\u5599\uff1b\u5b98\u65b9\u9875\u9762\u540c\u65f6\u516c\u5f00Apache-2.0\u8f6f\u4ef6\u6808\u3001MuJoCo\u4eff\u771f\u3001\u5f3a\u5316\u5b66\u4e60\u8bad\u7ec3\u6d41\u7a0b\u548c7\u4e2a\u9884\u8bad\u7ec3\u52a8\u4f5c\u7b56\u7565\u3002\u73b0\u9636\u6bb5\u786e\u8ba4\u7684\u662f\u9884\u8ba2\u3001\u4ee3\u7801\u4e0e\u6f14\u793a\u5f00\u653e\uff0c\u4e0d\u662f\u6279\u91cf\u4ea4\u4ed8\uff1b\u76f8\u673a\u3001LiDAR\u3001\u65e0\u7ebf\u89c4\u683c\u3001SDK\u8bed\u8a00\u548c\u5e74\u9f84\u5efa\u8bae\u4ecd\u6807\u6ce8\u4e3a\u672a\u6700\u7ec8\u786e\u5b9a\uff0c\u771f\u5b9e\u73af\u5883\u6027\u80fd\u4e5f\u6ca1\u6709\u72ec\u7acb\u957f\u671f\u6d4b\u8bd5\u3002<\/p>\n<p>\u6765\u6e90<\/p>\n<p>Anthropic\uff0cPreviewing the Model Hardware Standard\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/www.anthropic.com\/news\/model-hardware-standard-research-preview\" rel=\"nofollow\">https:\/\/www.anthropic.com\/news\/model-hardware-standard-research-preview<\/a><\/p>\n<p>Reuters\uff0cAnthropic unveils new framework allowing AI agents to operate physical devices\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/www.reuters.com\/technology\/anthropic-unveils-new-framework-allowing-ai-agents-operate-physical-devices-2026-08-27\/\" rel=\"nofollow\">https:\/\/www.reuters.com\/technology\/anthropic-unveils-new-framework-allowing-ai-agents-operate-physical-devices-2026-08-27\/<\/a><\/p>\n<p>QuEra Computing\uff0cHolding the Light: Teaching an AI to Lock and Tune our Quantum Computer&#39;s Lasers\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/www.quera.com\/blog-posts\/holding-the-light-teaching-an-ai-to-lock-and-tune-our-quantum-computers-lasers\" rel=\"nofollow\">https:\/\/www.quera.com\/blog-posts\/holding-the-light-teaching-an-ai-to-lock-and-tune-our-quantum-computers-lasers<\/a><\/p>\n<p>Google\uff0cGemini Omni 1.1 Flash lets you build with more control\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/blog.google\/innovation-and-ai\/technology\/developers-tools\/build-with-gemini-omni-1-1-flash\/\" rel=\"nofollow\">https:\/\/blog.google\/innovation-and-ai\/technology\/developers-tools\/build-with-gemini-omni-1-1-flash\/<\/a><\/p>\n<p>The Decoder\uff0cGoogle&#39;s Gemini Omni 1.1 Flash makes AI video generation cheaper and more flexible\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/the-decoder.com\/googles-gemini-omni-1-1-flash-makes-ai-video-generation-cheaper-and-more-flexible\/\" rel=\"nofollow\">https:\/\/the-decoder.com\/googles-gemini-omni-1-1-flash-makes-ai-video-generation-cheaper-and-more-flexible\/<\/a><\/p>\n<p>Google\uff0cIntelligent transcription with Gemini 3.5 Transcribe\uff0c2026\u5e748\u670826\u65e5<br \/><a href=\"https:\/\/blog.google\/innovation-and-ai\/models-and-research\/gemini-models\/gemini-3-5-transcribe\/\" rel=\"nofollow\">https:\/\/blog.google\/innovation-and-ai\/models-and-research\/gemini-models\/gemini-3-5-transcribe\/<\/a><\/p>\n<p>Artificial Analysis\uff0cSpeech to Text Providers Leaderboard\uff0c2026\u5e748\u670827\u65e5\u8bbf\u95ee<br \/><a href=\"https:\/\/artificialanalysis.ai\/speech-to-text\/non-streaming\" rel=\"nofollow\">https:\/\/artificialanalysis.ai\/speech-to-text\/non-streaming<\/a><\/p>\n<p>NVIDIA\uff0cDelivering Vera: NVIDIA&#39;s First CPU Built for Agents Is Shipping Now\uff0c2026\u5e748\u670827\u65e5\u66f4\u65b0<br \/><a href=\"https:\/\/blogs.nvidia.com\/blog\/vera-cpu-delivery\/\" rel=\"nofollow\">https:\/\/blogs.nvidia.com\/blog\/vera-cpu-delivery\/<\/a><\/p>\n<p>Amazon\uff0cAWS and NVIDIA expand partnership for next-gen AI infrastructure\uff0c2026\u5e748\u670826\u65e5<br \/><a href=\"https:\/\/www.aboutamazon.com\/news\/aws\/aws-nvidia-2-million-gpus-ai\" rel=\"nofollow\">https:\/\/www.aboutamazon.com\/news\/aws\/aws-nvidia-2-million-gpus-ai<\/a><\/p>\n<p>Reuters\uff0cNvidia begins Vera CPU sales pitch to Chinese clients, sources say\uff0c2026\u5e746\u670812\u65e5<br \/><a href=\"https:\/\/www.reuters.com\/world\/china\/nvidia-begins-vera-cpu-sales-pitch-chinese-clients-sources-say-2026-06-12\/\" rel=\"nofollow\">https:\/\/www.reuters.com\/world\/china\/nvidia-begins-vera-cpu-sales-pitch-chinese-clients-sources-say-2026-06-12\/<\/a><\/p>\n<p>Pollen Robotics\uff0cMicroduck\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/pollen-robotics.com\/microduck\/\" rel=\"nofollow\">https:\/\/pollen-robotics.com\/microduck\/<\/a><\/p>\n<p>The Verge\uff0cHugging Face&#39;s new robot is an adorable rollerskating duck\uff0c2026\u5e748\u670827\u65e5<br \/><a href=\"https:\/\/www.theverge.com\/gadgets\/985549\/hugging-face-microduck-robot\" rel=\"nofollow\">https:\/\/www.theverge.com\/gadgets\/985549\/hugging-face-microduck-robot<\/a><\/p>\n<p>","protected":false},"excerpt":{"rendered":"","protected":false},"author":17897162,"featured_media":0,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_coblocks_attr":"","_coblocks_dimensions":"","_coblocks_responsive_height":"","_coblocks_accordion_ie_support":"","_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":true,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_geoff_meta_description_en":"A verified update on Anthropic's Model Hardware Standard, Google's video and speech models, NVIDIA Vera CPU deliveries and the Microduck open-source robot.","_geoff_meta_description_zh":"\u6838\u9a8cAnthropic\u6a21\u578b\u786c\u4ef6\u6807\u51c6\u3001Google\u89c6\u9891\u4e0e\u8bed\u97f3\u6a21\u578b\u3001NVIDIA Vera CPU\u4ea4\u4ed8\u53caMicroduck\u5f00\u6e90\u673a\u5668\u4eba\u9884\u8ba2\u72b6\u6001\u3002","_wpcom_ai_launchpad_first_post":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"jetpack_post_was_ever_published":false},"categories":[787298266],"tags":[787295121,787296348,787298267,787296349,787298268,787295122,787294346,787298269,787298270,787294292],"class_list":["post-18549","post","type-post","status-publish","format-standard","hentry","category-ai-technology-observation","tag-ai-agents","tag-ai-research","tag-ai-technology","tag-ai-research-zh","tag-ai-technology-zh","tag-ai-agents-zh","tag-artificial-intelligence","tag-robotics","tag-robotics-zh","tag-787294292",""],"translatepress_titles":{"en_AU":"AI Technology Observations | 28 August 2026, 08:00","zh_CN":"AI\u6280\u672f\u89c2\u5bdf\uff5c2026\u5e748\u670828\u65e5 08\u65f6"},"jetpack_publicize_connections":[],"jetpack_likes_enabled":false,"jetpack_sharing_enabled":true,"jetpack_shortlink":"https:\/\/wp.me\/p1ar9H-4Pb","jetpack_featured_media_url":"","_links":{"self":[{"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/posts\/18549","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/users\/17897162"}],"replies":[{"embeddable":true,"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/comments?post=18549"}],"version-history":[{"count":0,"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/posts\/18549\/revisions"}],"wp:attachment":[{"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/media?parent=18549"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/categories?post=18549"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/geoffreychen.com\/zh\/wp-json\/wp\/v2\/tags?post=18549"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}