\u65e5\u671f\uff1a2026-05-07<\/p>\n
\u672c\u671f\u805a\u7126\uff1a\u91cd\u70b9\u5173\u6ce8\u6a21\u578b\u53d1\u5e03\u4e0e release notes\u3001\u5b98\u65b9 engineering blog\u3001AI coding \/ agent \/ SRE\u3001\u8bc4\u6d4b\u699c\u5355\u53d8\u5316\u3001\u5f00\u53d1\u8005\u5b9e\u8df5\u535a\u5ba2\u3001\u6846\u67b6\u751f\u6001\u3001\u5f00\u6e90\u6a21\u578b\u4e0e\u771f\u5b9e\u7528\u6237\u89c6\u89d2\uff1b\u5f53 HN\u3001Reddit\u3001Hugging Face \u7b49\u793e\u533a\u6e90\u53ef\u8bbf\u95ee\u65f6\u4f18\u5148\u7eb3\u5165\u3002<\/p>\n
Artificial Analysis \u6700\u65b0\u6a21\u578b\u6392\u540d\u89c2\u5bdf<\/strong>\uff08Artificial Analysis\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Artificial Analysis \u6700\u65b0\u6a21\u578b\u6392\u540d\u663e\u793a\uff0cGPT-5.5\uff08xhigh \u4e0e high \u7248\u672c\uff09\u5728\u667a\u80fd\u6307\u6570\u4e0a\u9886\u5148\uff0cClaude Opus 4.7 (max) \u4e0e Gemini 3.1 Pro Preview \u7d27\u968f\u5176\u540e\u3002\u8f93\u51fa\u901f\u5ea6\u65b9\u9762\uff0cMercury 2 \u4ee5 753 tokens\/\u79d2 \u5c45\u9996\uff0cQwen3.5 0.8B \u8fbe 359 tokens\/\u79d2\u3002\u5ef6\u8fdf\u6700\u4f4e\u7684\u662f Qwen3.5 4B\uff080.44\u79d2\uff09\u3002\u4ef7\u683c\u6700\u4eb2\u6c11\u7684\u6a21\u578b\u4e3a Qwen3.5 0.8B\uff08\u6bcf\u767e\u4e07 tokens \u4ec5 $0.02\uff09\u3002\u4e0a\u4e0b\u6587\u7a97\u53e3\u6700\u5927\u7684\u5219\u662f Llama 4 Scout\uff081000\u4e07 tokens\uff09\u4e0e Grok 4.20\uff08200\u4e07 tokens\uff09\u3002\u8be5\u6392\u540d\u57fa\u4e8e\u5305\u542b GDPval-AA\u3001Terminal-Bench Hard\u3001Humanity's Last Exam \u7b49 10 \u9879\u8bc4\u4f30\u7684 Intelligence Index v4.0\uff0c\u8986\u76d6 513 \u4e2a\u6a21\u578b\u3002<\/p>\n English Summary:<\/strong> Artificial Analysis' latest model rankings show GPT-5.5 (xhigh and high variants) leading in intelligence, followed by Claude Opus 4.7 (max) and Gemini 3.1 Pro Preview. Mercury 2 tops output speed at 753 tokens\/s, while Qwen3.5 4B offers the lowest latency at 0.44s. Qwen3.5 0.8B is the most affordable at $0.02 per million tokens. Llama 4 Scout features the largest context window at 10M tokens. Rankings are based on the Intelligence Index v4.0 covering 513 models across 10 evaluations including GDPval-AA, Terminal-Bench Hard, and Humanity's Last Exam.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Introducing Claude Opus 4.7<\/strong>\uff08Anthropic News\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Anthropic \u6b63\u5f0f\u53d1\u5e03 Claude Opus 4.7\uff0c\u5728\u9ad8\u7ea7\u8f6f\u4ef6\u5de5\u7a0b\u4efb\u52a1\u4e0a\u8f83 Opus 4.6 \u6709\u663e\u8457\u63d0\u5347\uff0c\u5c24\u5176\u5728\u5904\u7406\u6700\u56f0\u96be\u7684\u7f16\u7801\u4efb\u52a1\u65f6\u8868\u73b0\u51fa\u8272\u3002\u8be5\u6a21\u578b\u5177\u5907\u66f4\u9ad8\u5206\u8fa8\u7387\u7684\u89c6\u89c9\u80fd\u529b\uff08\u652f\u6301\u957f\u8fbe 2576 \u50cf\u7d20\u7684\u56fe\u50cf\uff09\uff0c\u5728\u4e13\u4e1a\u4efb\u52a1\u4e2d\u5c55\u73b0\u51fa\u66f4\u4f73\u7684\u5ba1\u7f8e\u4e0e\u521b\u9020\u529b\u3002Anthropic \u4e3a\u5176\u5f15\u5165\u4e86\u5b9e\u65f6\u7f51\u7edc\u5b89\u5168\u9632\u62a4\u673a\u5236\uff0c\u81ea\u52a8\u68c0\u6d4b\u5e76\u62e6\u622a\u9ad8\u98ce\u9669\u7f51\u7edc\u653b\u51fb\u8bf7\u6c42\uff0c\u540c\u65f6\u63a8\u51fa Cyber Verification Program \u4f9b\u5b89\u5168\u4e13\u4e1a\u4eba\u5458\u7533\u8bf7\u5408\u6cd5\u4f7f\u7528\u3002Opus 4.7 \u65b0\u589e xhigh \u52aa\u529b\u7ea7\u522b\uff0cClaude Code \u9ed8\u8ba4 effort \u63d0\u5347\u81f3 xhigh\u3002\u5b9a\u4ef7\u7ef4\u6301\u4e0d\u53d8\uff1a\u8f93\u5165 $5\/\u767e\u4e07 tokens\uff0c\u8f93\u51fa $25\/\u767e\u4e07 tokens\u3002\u591a\u5bb6\u5408\u4f5c\u4f19\u4f34\u5982 Replit\u3001Notion\u3001Cursor\u3001Vercel \u7b49\u5728\u5185\u90e8\u8bc4\u6d4b\u4e2d\u62a5\u544a\u4e86\u663e\u8457\u7684\u6027\u80fd\u63d0\u5347\u3002<\/p>\n English Summary:<\/strong> Anthropic officially released Claude Opus 4.7, featuring notable improvements in advanced software engineering over Opus 4.6, particularly on the most difficult coding tasks. The model offers enhanced vision capabilities supporting images up to 2,576 pixels on the long edge, and demonstrates better taste and creativity in professional tasks. Anthropic introduced real-time cyber safeguards that automatically detect and block high-risk cybersecurity requests, alongside a new Cyber Verification Program for legitimate security research. Opus 4.7 introduces a new xhigh effort level between high and max, with Claude Code defaulting to xhigh. Pricing remains at $5\/M input tokens and $25\/M output tokens. Partners including Replit, Notion, Cursor, and Vercel reported significant performance gains in internal evaluations.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Featured An update on recent Claude Code quality reports<\/strong>\uff08Anthropic Engineering\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Anthropic \u53d1\u5e03\u6280\u672f\u590d\u76d8\uff0c\u89e3\u91ca\u4e86\u8fc7\u53bb\u4e00\u4e2a\u6708 Claude Code\u3001Agent SDK \u548c Claude Cowork \u7528\u6237\u53cd\u9988\u8d28\u91cf\u4e0b\u964d\u7684\u4e09\u9879\u6839\u672c\u539f\u56e0\u3002\u7b2c\u4e00\u9879\u662f 3 \u6708 4 \u65e5\u5c06\u9ed8\u8ba4\u63a8\u7406\u52aa\u529b\u7ea7\u522b\u4ece high \u964d\u81f3 medium\uff0c\u5bfc\u81f4\u6a21\u578b\u667a\u80fd\u611f\u77e5\u4e0b\u964d\uff0c\u5df2\u4e8e 4 \u6708 7 \u65e5\u56de\u6eda\u3002\u7b2c\u4e8c\u9879\u662f 3 \u6708 26 \u65e5\u5f15\u5165\u7684\u7f13\u5b58\u4f18\u5316\u5b58\u5728 bug\uff0c\u5bfc\u81f4\u8d85\u8fc7\u4e00\u5c0f\u65f6\u7a7a\u95f2\u7684\u4f1a\u8bdd\u5728\u540e\u7eed\u6bcf\u4e00\u8f6e\u90fd\u4f1a\u4e22\u5931\u5386\u53f2\u63a8\u7406\u8bb0\u5f55\uff0c\u4f7f Claude \u663e\u5f97\u5065\u5fd8\u548c\u91cd\u590d\uff0c\u5df2\u4e8e 4 \u6708 10 \u65e5\u4fee\u590d\u3002\u7b2c\u4e09\u9879\u662f 4 \u6708 16 \u65e5\u6dfb\u52a0\u7684\u51cf\u5c11\u5197\u957f\u5ea6\u7684\u7cfb\u7edf\u63d0\u793a\u8bcd\u610f\u5916\u964d\u4f4e\u4e86\u7f16\u7801\u8d28\u91cf\uff0c\u5df2\u4e8e 4 \u6708 20 \u65e5\u56de\u6eda\u3002Anthropic \u5df2\u91cd\u7f6e\u6240\u6709\u8ba2\u9605\u8005\u7684\u4f7f\u7528\u989d\u5ea6\uff0c\u5e76\u627f\u8bfa\u6539\u8fdb\u5185\u90e8\u6d4b\u8bd5\u6d41\u7a0b\uff0c\u5305\u62ec\u8ba9\u66f4\u591a\u5458\u5de5\u4f7f\u7528\u516c\u5171\u7248\u672c\u3001\u589e\u5f3a Code Review \u5de5\u5177\uff0c\u4ee5\u53ca\u5bf9\u7cfb\u7edf\u63d0\u793a\u8bcd\u53d8\u66f4\u5b9e\u65bd\u66f4\u4e25\u683c\u7684\u8bc4\u4f30\u63a7\u5236\u3002<\/p>\n English Summary:<\/strong> Anthropic published a technical postmortem explaining three root causes of reported quality degradation in Claude Code, Agent SDK, and Claude Cowork over the past month. First, on March 4, the default reasoning effort was changed from high to medium, reducing perceived intelligence\u2014reverted on April 7. Second, a March 26 caching optimization bug caused sessions idle over an hour to drop historical reasoning on every subsequent turn, making Claude appear forgetful\u2014fixed on April 10. Third, an April 16 system prompt change to reduce verbosity inadvertently hurt coding quality\u2014reverted on April 20. Anthropic has reset usage limits for all subscribers and committed to improving internal testing, including broader staff use of public builds, enhanced Code Review tools, and stricter evaluation controls for system prompt changes.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Scaling Managed Agents: Decoupling the brain from the hands<\/strong>\uff08Anthropic Engineering\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Anthropic \u5de5\u7a0b\u56e2\u961f\u53d1\u5e03\u300aScaling Managed Agents: Decoupling the brain from the hands\u300b\uff0c\u9610\u8ff0\u5176\u6258\u7ba1\u4ee3\u7406\u670d\u52a1\u7684\u67b6\u6784\u8bbe\u8ba1\u54f2\u5b66\u3002\u8be5\u670d\u52a1\u901a\u8fc7\u5c06\u4ee3\u7406\u7ec4\u4ef6\u865a\u62df\u5316\u4e3a\u4e09\u5927\u62bd\u8c61\u63a5\u53e3\u2014\u2014Session\uff08\u4e8b\u4ef6\u65e5\u5fd7\uff09\u3001Harness\uff08\u8c03\u7528 Claude \u7684\u5faa\u73af\uff09\u548c Sandbox\uff08\u4ee3\u7801\u6267\u884c\u73af\u5883\uff09\u2014\u2014\u5b9e\u73b0"\u5927\u8111"\uff08\u6a21\u578b\u4e0e harness\uff09\u4e0e"\u53cc\u624b"\uff08\u6c99\u7bb1\u4e0e\u5de5\u5177\uff09\u7684\u89e3\u8026\u3002\u8fd9\u79cd\u8bbe\u8ba1\u4f7f\u5404\u7ec4\u4ef6\u53ef\u72ec\u7acb\u5931\u8d25\u548c\u66ff\u6362\uff0c\u907f\u514d\u5355\u70b9\u6545\u969c\u3002\u89e3\u8026\u540e\uff0cp50 \u9996 token \u5ef6\u8fdf\u964d\u4f4e\u7ea6 60%\uff0cp95 \u964d\u4f4e\u8d85 90%\u3002\u6587\u7ae0\u8fd8\u8ba8\u8bba\u4e86\u5b89\u5168\u8fb9\u754c\u8bbe\u8ba1\uff1a\u901a\u8fc7\u5c06\u51ed\u8bc1\u5b58\u50a8\u5728\u6c99\u7bb1\u5916\u90e8\u7684 vault \u4e2d\uff0c\u5e76\u901a\u8fc7\u4ee3\u7406\u8fdb\u884c\u5de5\u5177\u8c03\u7528\uff0c\u9632\u6b62\u63d0\u793a\u8bcd\u6ce8\u5165\u653b\u51fb\u83b7\u53d6\u654f\u611f\u4ee4\u724c\u3002\u8be5\u67b6\u6784\u652f\u6301\u591a\u5927\u8111\u3001\u591a\u53cc\u624b\u7684\u7075\u6d3b\u7ec4\u5408\uff0c\u5141\u8bb8\u4ee3\u7406\u6839\u636e\u4efb\u52a1\u9700\u6c42\u52a8\u6001\u8fde\u63a5\u4e0d\u540c\u7684\u6267\u884c\u73af\u5883\u3002<\/p>\n English Summary:<\/strong> Anthropic's engineering team published "Scaling Managed Agents: Decoupling the brain from the hands," detailing the architectural design philosophy behind their Managed Agents service. The system virtualizes agent components into three abstractions\u2014Session (event log), Harness (loop calling Claude), and Sandbox (execution environment)\u2014decoupling the "brain" (model and harness) from the "hands" (sandboxes and tools). This design allows components to fail and be replaced independently, eliminating single points of failure. Decoupling reduced p50 time-to-first-token latency by roughly 60% and p95 by over 90%. The article discusses security boundaries: credentials are stored in a vault outside the sandbox, with tools called via a proxy to prevent prompt injection attacks from accessing sensitive tokens. The architecture supports flexible combinations of multiple brains and hands, allowing agents to dynamically connect to different execution environments based on task requirements.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Barry Diller trusts Sam Altman. But \u2018trust is irrelevant\u2019 as AGI nears, he says.<\/strong>\uff08TechCrunch AI\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>\u4ebf\u4e07\u5bcc\u7fc1\u3001IAC \u4e0e Expedia \u96c6\u56e2\u8463\u4e8b\u957f Barry Diller \u5728\u300a\u534e\u5c14\u8857\u65e5\u62a5\u300b"Future of Everything" \u5927\u4f1a\u4e0a\u4e3a OpenAI CEO Sam Altman \u8fa9\u62a4\uff0c\u8868\u793a\u4ed6\u76f8\u4fe1 Altman \u662f\u771f\u8bda\u4e14\u4ef7\u503c\u89c2\u7aef\u6b63\u7684\u4eba\u3002\u7136\u800c\uff0cDiller \u5f3a\u8c03\u968f\u7740\u901a\u7528\u4eba\u5de5\u667a\u80fd\uff08AGI\uff09\u4e34\u8fd1\uff0c"\u4fe1\u4efb\u5df2\u65e0\u5173\u7d27\u8981"\uff0c\u56e0\u4e3a AI \u53d1\u5c55\u7684\u540e\u679c\u8fde\u521b\u9020\u8005\u672c\u8eab\u4e5f\u65e0\u6cd5\u9884\u6d4b\u3002\u4ed6\u6307\u51fa\uff0cAI \u9886\u57df\u7684\u521b\u9020\u8005\u4eec\u90fd\u5bf9\u6280\u672f\u8fdb\u5c55\u611f\u5230"\u60ca\u5947"\uff0c\u8fd9\u610f\u5473\u7740\u6280\u672f\u6f14\u8fdb\u5b58\u5728\u5de8\u5927\u7684\u4e0d\u786e\u5b9a\u6027\u3002Diller \u8b66\u544a\u79f0\uff0c\u4eba\u7c7b\u5fc5\u987b\u4e3a AGI \u8bbe\u7acb\u9632\u62a4\u680f\uff08guardrails\uff09\uff0c\u5426\u5219"\u53e6\u4e00\u79cd\u529b\u91cf\u2014\u2014AGI \u672c\u8eab\u2014\u2014\u5c06\u4f1a\u81ea\u884c\u51b3\u5b9a"\uff0c\u4e00\u65e6\u91ca\u653e\u4fbf\u65e0\u6cd5\u56de\u5934\u3002\u4ed6\u8ba4\u4e3a AI \u5c06\u6539\u53d8\u51e0\u4e4e\u6240\u6709\u4e8b\u7269\uff0c\u5c3d\u7ba1\u5bf9\u5de8\u989d\u6295\u8d44\u80fd\u5426\u5151\u73b0\u6301\u6000\u7591\u6001\u5ea6\uff0c\u4f46\u6280\u672f\u8fdb\u6b65\u5fc5\u5c06\u6301\u7eed\u3002<\/p>\n English Summary:<\/strong> Billionaire media mogul Barry Diller, chairman of IAC and Expedia Group, defended OpenAI CEO Sam Altman at The Wall Street Journal's "Future of Everything" conference, stating he believes Altman is sincere and has good values. However, Diller emphasized that as artificial general intelligence (AGI) nears, "trust is irrelevant" because the consequences of AI development are unpredictable even to its creators. He noted that AI developers themselves express "wonder" at what they're creating, indicating vast uncertainty in technological evolution. Diller warned that humans must establish guardrails for AGI, or "another force, an AGI force, will do it themselves," with no going back once unleashed. He believes AI will change almost everything, expressing skepticism about whether massive investments will pay off, but asserting that progress will continue regardless.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Validating agentic behavior when \u201ccorrect\u201d isn\u2019t deterministic<\/strong>\uff08GitHub AI\/ML\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>GitHub \u53d1\u5e03\u4e86\u4e00\u7bc7\u5173\u4e8e\u9a8c\u8bc1 AI Agent \u884c\u4e3a\u7684\u5de5\u7a0b\u535a\u5ba2\uff0c\u63d0\u51fa\u4e86\u4e00\u79cd\u57fa\u4e8e\u652f\u914d\u5206\u6790\uff08Dominator Analysis\uff09\u7684"\u4fe1\u4efb\u5c42"\u6846\u67b6\uff0c\u7528\u4e8e\u89e3\u51b3\u975e\u786e\u5b9a\u6027 Agent \u884c\u4e3a\u7684\u9a8c\u8bc1\u96be\u9898\u3002\u4f20\u7edf\u6d4b\u8bd5\u5047\u8bbe\u884c\u4e3a\u662f\u53ef\u91cd\u590d\u7684\uff0c\u4f46 Agent \u5728\u771f\u5b9e\u73af\u5883\uff08\u5982 UI\u3001\u6d4f\u89c8\u5668\uff09\u4e2d\u6267\u884c\u65f6\uff0c\u52a0\u8f7d\u65f6\u95f4\u3001\u7f51\u7edc\u5ef6\u8fdf\u7b49\u56e0\u7d20\u4f1a\u5bfc\u81f4\u6267\u884c\u8def\u5f84\u591a\u53d8\u3002\u8be5\u6846\u67b6\u5c06\u6267\u884c\u8f68\u8ff9\u5efa\u6a21\u4e3a\u56fe\u7ed3\u6784\u800c\u975e\u7ebf\u6027\u811a\u672c\uff0c\u901a\u8fc7\u524d\u7f00\u6811\u63a5\u53d7\u5668\uff08PTA\uff09\u5408\u5e76\u591a\u6b21\u6210\u529f\u6267\u884c\uff0c\u518d\u5229\u7528\u7f16\u8bd1\u5668\u7406\u8bba\u4e2d\u7684\u652f\u914d\u5206\u6790\u63d0\u53d6"\u5fc5\u987b\u5b8c\u6210\u7684\u5173\u952e\u8282\u70b9"\uff0c\u4ece\u800c\u533a\u5206\u5fc5\u8981\u884c\u4e3a\u4e0e\u5076\u53d1\u566a\u97f3\u3002\u5b9e\u9a8c\u8868\u660e\uff0c\u8be5\u65b9\u6cd5\u5728 VS Code \u73af\u5883\u4e2d\u8fbe\u5230\u4e86 100% \u7684\u51c6\u786e\u7387\uff0c\u663e\u8457\u4f18\u4e8e Agent \u81ea\u8bc4\u4f30\u7684 82.2%\u3002\u8fd9\u4e00\u65b9\u6cd5\u4e3a CI \u6d41\u6c34\u7ebf\u4e2d\u96c6\u6210 Agent \u6d4b\u8bd5\u63d0\u4f9b\u4e86\u53ef\u89e3\u91ca\u3001\u8f7b\u91cf\u4e14\u9c81\u68d2\u7684\u89e3\u51b3\u65b9\u6848\u3002<\/p>\n English Summary:<\/strong> GitHub published an engineering blog on validating AI agent behavior, introducing a "Trust Layer" framework based on dominator analysis to address non-deterministic agent validation. Traditional testing assumes repeatable behavior, but agents operating in real environments (UIs, browsers) face variability from loading times and network latency. The framework models execution traces as graphs using Prefix Tree Acceptors (PTA), merging successful runs and applying compiler-theory dominator analysis to extract essential milestones versus incidental noise. Experiments in VS Code achieved 100% accuracy, significantly outperforming agent self-assessment at 82.2%. This provides an explainable, lightweight, and robust solution for integrating agent testing into CI pipelines.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Cost effective deployment of vision-language models for pet behavior detection on AWS Inferentia2<\/strong>\uff08AWS ML Blog\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>AWS \u673a\u5668\u5b66\u4e60\u535a\u5ba2\u4ecb\u7ecd\u4e86\u53f0\u6e7e\u5ba0\u7269\u79d1\u6280\u516c\u53f8 Tomofun\uff08Furbo \u667a\u80fd\u6444\u50cf\u5934\u5f00\u53d1\u5546\uff09\u5982\u4f55\u5c06\u5176\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\uff08VLM\uff09\u63a8\u7406\u5de5\u4f5c\u8d1f\u8f7d\u4ece GPU \u8fc1\u79fb\u81f3 AWS Inferentia2\uff08Inf2\uff09\u5b9e\u4f8b\uff0c\u5b9e\u73b0\u6210\u672c\u4f18\u5316\u3002Furbo \u9700\u8981\u5b9e\u65f6\u68c0\u6d4b\u5ba0\u7269\u884c\u4e3a\uff08\u5982\u5420\u53eb\u3001\u5954\u8dd1\uff09\uff0c\u539f\u6709 GPU \u65b9\u6848\u6210\u672c\u9ad8\u6602\u3002\u901a\u8fc7\u4f7f\u7528 Neuron SDK \u5c06 BLIP \u6a21\u578b\u7684\u56fe\u50cf\u7f16\u7801\u5668\u3001\u6587\u672c\u7f16\u7801\u5668\u548c\u6587\u672c\u89e3\u7801\u5668\u5206\u522b\u7f16\u8bd1\u4e3a Neuron \u4f18\u5316\u683c\u5f0f\uff0c\u5e76\u91c7\u7528\u8f7b\u91cf\u7ea7 Wrapper \u7c7b\u9002\u914d I\/O \u63a5\u53e3\uff0cTomofun \u5728\u4fdd\u6301\u539f\u59cb\u6a21\u578b\u903b\u8f91\u4e0d\u53d8\u7684\u60c5\u51b5\u4e0b\u5b8c\u6210\u4e86\u8fc1\u79fb\u3002\u538b\u529b\u6d4b\u8bd5\u9a8c\u8bc1\u4e86 Inf2 \u5b9e\u4f8b\u53ef\u540c\u65f6\u5904\u7406\u6570\u5341\u4e07\u53f0\u8bbe\u5907\u7684\u5e76\u53d1\u8bf7\u6c42\u3002\u6700\u7ec8\uff0cTomofun \u5b9e\u73b0\u4e86 83% \u7684\u6210\u672c\u964d\u4f4e\uff0c\u540c\u65f6\u7ef4\u6301\u4e86\u4f4e\u5ef6\u8fdf\u548c\u9ad8\u541e\u5410\u91cf\u7684\u5b9e\u65f6\u63a8\u7406\u6027\u80fd\u3002<\/p>\n English Summary:<\/strong> AWS ML Blog details how Taiwan-based pet-tech startup Tomofun (maker of Furbo smart cameras) migrated vision-language model (VLM) inference from GPUs to AWS Inferentia2 (Inf2) instances for cost optimization. Furbo requires real-time pet behavior detection (barking, running), and the original GPU solution was costly. Using the Neuron SDK, Tomofun compiled BLIP model components (image encoder, text encoder, text decoder) into Neuron-optimized formats with lightweight wrapper classes adapting I\/O interfaces\u2014keeping original model logic unchanged. Stress testing validated Inf2's ability to handle concurrent requests across hundreds of thousands of devices. The migration achieved an 83% cost reduction while maintaining low-latency, high-throughput real-time inference.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n LinkedIn Consolidates Hiring Data Pipelines to Power AI Driven Talent Systems<\/strong>\uff08InfoQ AI\/ML\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>LinkedIn \u5de5\u7a0b\u56e2\u961f\u4ecb\u7ecd\u4e86\u5176\u7edf\u4e00\u62db\u8058\u6570\u636e\u96c6\u6210\u5e73\u53f0\uff0c\u65e8\u5728\u89e3\u51b3\u62db\u8058\u6570\u636e\u5206\u6563\u3001\u683c\u5f0f\u4e0d\u4e00\u81f4\u7684\u95ee\u9898\u3002\u8be5\u5e73\u53f0\u901a\u8fc7\u6807\u51c6\u5316\u5c42\uff08Standardization\uff09\u3001\u7f16\u6392\u5c42\uff08Orchestration\uff09\u548c\u589e\u5f3a\u5c42\uff08Enhancement\uff09\u4e09\u5c42\u67b6\u6784\uff0c\u5c06\u6765\u81ea ATS\u3001\u62db\u8058\u7f51\u7ad9\u7b49\u5f02\u6784\u6570\u636e\u6e90\u7684\u6570\u636e\u7edf\u4e00\u6574\u5408\u3002\u5e95\u5c42\u91c7\u7528 Temporal \u7f16\u6392\u5de5\u4f5c\u6d41\u3001Kafka \u6d41\u5904\u7406\u548c Espresso \u5b58\u50a8\uff0c\u652f\u6301\u53ef\u91cd\u653e\u7684\u53cc\u5411\u540c\u6b65\u3002\u8be5\u65b9\u6848\u4f7f\u5408\u4f5c\u4f19\u4f34\u4e0a\u7ebf\u65f6\u95f4\u7f29\u77ed 72%\uff0c\u6570\u636e\u8986\u76d6\u7387\u548c\u5b8c\u6574\u6027\u663e\u8457\u63d0\u5347\u3002\u7edf\u4e00\u7684\u6570\u636e\u57fa\u7840\u4e3a AI \u9a71\u52a8\u7684\u62db\u8058\u52a9\u624b\uff08Hiring Assistant\uff09\u63d0\u4f9b\u4e86\u611f\u77e5\u4e0e\u884c\u52a8\u63a5\u53e3\uff0c\u4f7f\u5176\u80fd\u591f\u8de8\u5019\u9009\u4eba\u6863\u6848\u3001\u804c\u4f4d\u8981\u6c42\u548c\u62db\u8058\u5b98\u4e92\u52a8\u8fdb\u884c\u667a\u80fd\u63a8\u8350\u548c\u51b3\u7b56\u652f\u6301\u3002<\/p>\n English Summary:<\/strong> LinkedIn's engineering team introduced a unified hiring data integration platform to address fragmented recruiting data and inconsistent schemas. The three-layer architecture\u2014Standardization, Orchestration, and Enhancement\u2014unifies data from heterogeneous sources like ATS and job boards. The underlying infrastructure uses Temporal-orchestrated workflows, Kafka streams, and Espresso storage for replayable bidirectional sync. This approach reduced partner onboarding time by 72% while improving data coverage and completeness. The standardized data foundation enables AI-driven hiring features through a perception and action interface for the Hiring Assistant, allowing intelligent recommendations across candidate profiles, job requirements, and recruiter interactions.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n [AINews] Silicon Valley gets Serious about Services<\/strong>\uff08Latent Space\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Latent Space \u7684 AI News \u5206\u6790\u4e86\u7845\u8c37 AI \u516c\u53f8\u5411\u670d\u52a1\u9886\u57df\u6269\u5c55\u7684\u8d8b\u52bf\u3002Anthropic \u4e0e Blackstone\u3001Hellman & Friedman\u3001Goldman Sachs \u6210\u7acb\u5408\u8d44\u4f01\u4e1a\uff0815 \u4ebf\u7f8e\u5143\uff09\uff0cOpenAI \u5219\u63a8\u51fa The Deployment Company\uff08\u7ea6 40 \u4ebf\u7f8e\u5143\u878d\u8d44\uff0c\u6295\u524d\u4f30\u503c 100 \u4ebf\u7f8e\u5143\uff09\uff0c\u4e24\u8005\u5747\u81f4\u529b\u4e8e\u4e3a\u4f01\u4e1a\u63d0\u4f9b\u5b9a\u5236\u5316\u7684 AI \u5b9e\u65bd\u670d\u52a1\u3002\u6587\u7ae0\u6307\u51fa\uff0c\u6a21\u578b\u5b9e\u9a8c\u5ba4\u6b63\u901a\u8fc7\u670d\u52a1\u4e1a\u52a1\u83b7\u53d6"\u6700\u540e\u4e00\u516c\u91cc"\u6536\u5165\u548c\u5dee\u5f02\u5316\u6570\u636e\uff0c\u56e0\u4e3a\u5c06 AI \u80fd\u529b\u7a33\u5b9a\u5e94\u7528\u4e8e\u4f01\u4e1a\u4e1a\u52a1\u6d41\u7a0b\u9700\u8981\u5927\u91cf\u7cfb\u7edf\u96c6\u6210\u3001\u5de5\u4f5c\u6d41\u6539\u9020\u548c\u53d8\u9769\u7ba1\u7406\u5de5\u4f5c\u3002\u4e0e\u6b64\u540c\u65f6\uff0cOpenAI \u53d1\u5e03\u4e86 GPT-5.5 Instant \u4f5c\u4e3a ChatGPT \u65b0\u9ed8\u8ba4\u6a21\u578b\uff0c\u5e76\u63a8\u51fa TypeScript \u7248 Agents SDK\uff1bGoogle \u53d1\u5e03 Gemma 4 \u591a token \u9884\u6d4b\u8349\u7a3f\u6a21\u578b\uff0c\u5ba3\u79f0\u89e3\u7801\u901f\u5ea6\u63d0\u5347 3 \u500d\uff1bRadixArk \u5b8c\u6210 1 \u4ebf\u7f8e\u5143\u79cd\u5b50\u8f6e\u878d\u8d44\uff0c\u57fa\u4e8e SGLang \u6784\u5efa\u5f00\u6e90\u63a8\u7406\u57fa\u7840\u8bbe\u65bd\u3002<\/p>\n English Summary:<\/strong> Latent Space's AI News analyzes the trend of Silicon Valley AI companies expanding into services. Anthropic formed a joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs ($1.5B), while OpenAI launched The Deployment Company (~$4B raised at $10B pre-money valuation)\u2014both focused on customized AI implementation for enterprises. Model labs are pursuing services for "last mile" revenue and differentiated data, as applying AI to business processes requires extensive system integration, workflow modernization, and change management. Meanwhile, OpenAI released GPT-5.5 Instant as ChatGPT's new default and a TypeScript Agents SDK; Google launched Gemma 4 multi-token prediction drafters claiming 3\u00d7 speedups; and RadixArk raised $100M seed funding to build open-source inference infrastructure on SGLang.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Uber uses OpenAI to help people earn smarter and book faster<\/strong>\uff08OpenAI News\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>OpenAI \u53d1\u5e03\u6848\u4f8b\u7814\u7a76\uff0c\u4ecb\u7ecd Uber \u5982\u4f55\u5229\u7528\u5176\u5927\u8bed\u8a00\u6a21\u578b\u548c\u5b9e\u65f6 API \u6784\u5efa AI \u52a9\u624b\u548c\u8bed\u97f3\u529f\u80fd\u3002Uber Assistant \u901a\u8fc7\u591a Agent \u67b6\u6784\u4e3a\u53f8\u673a\u63d0\u4f9b\u5b9e\u65f6\u5e02\u573a\u6d1e\u5bdf\u548c\u6536\u5165\u4f18\u5316\u5efa\u8bae\uff0c\u5c06\u590d\u6742\u7684\u6536\u76ca\u8d8b\u52bf\u548c\u70ed\u56fe\u6570\u636e\u8f6c\u5316\u4e3a\u53ef\u64cd\u4f5c\u7684\u5b9a\u4f4d\u5efa\u8bae\u3002\u7cfb\u7edf\u91c7\u7528\u5206\u5c42\u6a21\u578b\u7b56\u7565\uff08\u8f7b\u91cf\u7ea7\u6a21\u578b\u5904\u7406\u5206\u7c7b\u548c\u5feb\u901f\u54cd\u5e94\uff0c\u5927\u6a21\u578b\u5904\u7406\u590d\u6742\u4efb\u52a1\uff09\uff0c\u5e76\u901a\u8fc7 AI Guard \u5185\u90e8\u6cbb\u7406\u5c42\u786e\u4fdd\u5b89\u5168\u6027\u3001\u9690\u79c1\u548c\u4e00\u81f4\u6027\u3002\u8bed\u97f3\u529f\u80fd\u57fa\u4e8e OpenAI Realtime API\uff0c\u5141\u8bb8\u7528\u6237\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u5b8c\u6210\u53eb\u8f66\uff0c\u7cfb\u7edf\u53ef\u7406\u89e3\u5b8c\u6574\u610f\u56fe\u5e76\u540c\u6b65\u8bed\u97f3\u4e0e\u89c6\u89c9\u54cd\u5e94\u3002\u8be5\u52a9\u624b\u5df2\u5728\u7f8e\u56fd\u6570\u5341\u4e07\u53f8\u673a\u4e2d\u5b9e\u9a8c\u6027\u63a8\u51fa\uff0c\u5e2e\u52a9\u65b0\u53f8\u673a\u66f4\u5feb\u4e0a\u624b\uff0c\u540c\u65f6\u63d0\u5347\u8001\u53f8\u673a\u7684\u5e73\u53f0\u65f6\u95f4\u5229\u7528\u7387\u3002<\/p>\n English Summary:<\/strong> OpenAI published a case study on how Uber leverages its LLMs and Realtime API to build AI assistants and voice features. Uber Assistant uses a multi-agent architecture to provide drivers with real-time marketplace insights and earnings optimization, transforming complex data into actionable positioning recommendations. The system employs a tiered model strategy (lightweight models for classification\/fast responses, larger models for complex tasks) with an AI Guard governance layer for safety, privacy, and consistency. Voice features built on OpenAI's Realtime API allow users to book rides via natural speech, understanding full intent while synchronizing spoken and visual responses. The assistant has rolled out experimentally to hundreds of thousands of U.S. drivers, accelerating onboarding for new drivers and improving time utilization for experienced ones.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n How frontier enterprises are building an AI advantage<\/strong>\uff08OpenAI News\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>OpenAI \u53d1\u5e03 B2B Signals \u7814\u7a76\u62a5\u544a\uff0c\u63ed\u793a\u524d\u6cbf\u4f01\u4e1a\u5982\u4f55\u901a\u8fc7\u6df1\u5ea6 AI \u91c7\u7528\u6784\u5efa\u7ade\u4e89\u4f18\u52bf\u3002\u6570\u636e\u663e\u793a\uff0c\u5904\u4e8e 95 \u767e\u5206\u4f4d\u7684\u524d\u6cbf\u4f01\u4e1a\u6bcf\u540d\u5458\u5de5\u7684 AI \u4f7f\u7528\u91cf\u5df2\u8fbe\u5230\u666e\u901a\u4f01\u4e1a\u7684 3.5 \u500d\uff08\u4e00\u5e74\u524d\u4e3a 2 \u500d\uff09\uff0c\u5176\u4e2d\u6d88\u606f\u91cf\u4ec5\u5360\u5dee\u8ddd\u7684 36%\uff0c\u5927\u90e8\u5206\u5dee\u8ddd\u6e90\u4e8e\u66f4\u590d\u6742\u3001\u66f4\u6df1\u5165\u7684 AI \u5e94\u7528\u3002\u7279\u522b\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u4ee3\u7406\u5f0f\u5de5\u4f5c\u6d41\u6210\u4e3a\u6210\u719f\u5ea6\u7684\u65b0\u6807\u5fd7\u2014\u2014\u524d\u6cbf\u4f01\u4e1a\u5728 Codex \u4e0a\u7684\u6d88\u606f\u91cf\u662f\u666e\u901a\u4f01\u4e1a\u7684 16 \u500d\u3002\u62a5\u544a\u8fd8\u6307\u51fa\uff0cAI \u5e94\u7528\u6b63\u4ece\u901a\u7528\u751f\u4ea7\u529b\u5de5\u5177\u5411\u5404\u804c\u80fd\u6838\u5fc3\u4e1a\u52a1\u6e17\u900f\uff0cIT \u4e0e\u5b89\u5168\u56e2\u961f\u96c6\u4e2d\u4e8e\u64cd\u4f5c\u6307\u5357\u67e5\u8be2\uff0c\u8f6f\u4ef6\u5f00\u53d1\u56e2\u961f\u805a\u7126\u7f16\u7801\u4efb\u52a1\uff0c\u8d22\u52a1\u56e2\u961f\u5219\u7528\u4e8e\u5206\u6790\u8ba1\u7b97\u3002OpenAI \u5efa\u8bae\u4f01\u4e1a\u901a\u8fc7\u6d4b\u91cf\u4f7f\u7528\u6df1\u5ea6\u3001\u5efa\u7acb\u751f\u4ea7\u7ea7\u6cbb\u7406\u3001\u6295\u8d44\u8d4b\u80fd\u57f9\u8bad\u3001\u8bc6\u522b\u5e76\u63a8\u5e7f\u524d\u6cbf\u56e2\u961f\u7ecf\u9a8c\u3001\u4ece\u5bf9\u8bdd\u5f0f\u8f85\u52a9\u8f6c\u5411\u4ee3\u7406\u5f0f\u59d4\u6d3e\u5de5\u4f5c\u7b49\u65b9\u5f0f\u5411\u524d\u6cbf\u9760\u62e2\u3002<\/p>\n English Summary:<\/strong> OpenAI's B2B Signals research reveals how frontier enterprises build AI advantage through deeper adoption. Frontier firms (95th percentile) now use 3.5x more AI intelligence per worker than typical firms, up from 2x a year ago. Message volume accounts for only 36% of this gap; most stems from richer, more complex AI use. Agentic workflows mark the new maturity frontier, with leading firms sending 16x more Codex messages per worker. AI use is broadening across functions\u2014IT\/Security focuses on procedural guidance, Software Development on coding, and Finance on analysis. OpenAI recommends measuring depth of use, building production governance, investing in enablement, scaling frontier team practices, and moving from chat-based assistance to delegated agentic work.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n 🔬Doing Vibe Physics \u2014 Alex Lupsasca, OpenAI<\/strong>\uff08Latent Space\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Latent Space \u64ad\u5ba2\u6df1\u5ea6\u8bbf\u8c08 OpenAI \u7269\u7406\u5b66\u5bb6 Alex Lupsasca\uff0c\u8bb2\u8ff0 GPT-5.x \u5728\u7406\u8bba\u7269\u7406\u4e0e\u91cf\u5b50\u5f15\u529b\u9886\u57df\u53d6\u5f97\u7a81\u7834\u6027\u65b0\u6210\u679c\u7684\u6545\u4e8b\u3002Lupsasca \u662f 2024 \u5e74\u57fa\u7840\u7269\u7406\u65b0\u89c6\u91ce\u7a81\u7834\u5956\uff08\u88ab\u8a89\u4e3a"\u7269\u7406\u5965\u65af\u5361"\uff09\u5f97\u4e3b\uff0c\u4ed6\u56de\u5fc6 GPT-5 \u53d1\u5e03\u65f6\u80fd\u5728 30 \u5206\u949f\u5185\u590d\u73b0\u4ed6\u8017\u65f6\u6781\u957f\u5b8c\u6210\u7684\u4e00\u7bc7\u9876\u7ea7\u8bba\u6587\uff0c\u800c\u5f53\u65f6\u7684\u516c\u4f17\u53cd\u5e94\u5374\u76f8\u5bf9\u51b7\u6de1\u2014\u2014\u56e0\u4e3a\u6a21\u578b\u5728\u5199\u90ae\u4ef6\u7b49\u65e5\u5e38\u4efb\u52a1\u4e0a\u63d0\u5347\u6709\u9650\u3002\u5728\u7814\u7a76\u4e2d\uff0c\u56e2\u961f\u5411 ChatGPT \u63d0\u51fa\u4e00\u4e2a\u56f0\u6270\u4e13\u5bb6\u4e00\u5e74\u591a\u7684\u5173\u4e8e"\u5355\u8d1f\u80f6\u5b50\u6811\u632f\u5e45"\u7684\u96be\u9898\uff0c\u7ed3\u679c\u6a21\u578b\u5728\u6559\u6388\u62b5\u8fbe OpenAI \u4e4b\u524d\uff08\u751a\u81f3\u98de\u673a\u964d\u843d\u524d\uff09\u5c31\u5b8c\u5168\u89e3\u51b3\u4e86\u95ee\u9898\uff0c\u5e76\u53d1\u73b0\u4e86\u4e00\u4e2a\u7b80\u5316\u590d\u6742\u7ed3\u679c\u7684"\u534a\u5171\u7ebf\u6781\u9650"\u3002\u968f\u540e\u56e2\u961f\u8ba9\u6a21\u578b\u7814\u7a76\u5f15\u529b\u5b50\u95ee\u9898\uff0cChatGPT \u5728\u4e00\u5929\u5185\u8f93\u51fa\u4e86 110 \u9875\u5168\u65b0\u7684\u7269\u7406\u8ba1\u7b97\u3001\u65b0\u65b9\u6cd5\u548c\u65b0\u6280\u672f\uff0c\u6700\u7ec8\u5f62\u6210\u4e00\u7bc7\u7ecf\u4e09\u5468\u9a8c\u8bc1\u7684\u91cf\u5b50\u5f15\u529b\u65b0\u6210\u679c\u8bba\u6587\u3002Lupsasca \u5c06\u8fd9\u79cd\u7814\u7a76\u65b9\u5f0f\u79f0\u4e3a"Vibe Physics"\u2014\u2014\u4e0e Vibe Coding \u4e0d\u540c\uff0c\u8fd9\u771f\u6b63\u6269\u5c55\u4e86\u4eba\u7c7b\u77e5\u8bc6\u7684\u524d\u6cbf\u8fb9\u754c\u3002<\/p>\n English Summary:<\/strong> Latent Space podcast features Alex Lupsasca, OpenAI physicist and 2024 New Horizons Breakthrough Prize winner, on how GPT-5.x derived novel results in theoretical physics and quantum gravity. While public reception to GPT-5 was lukewarm for everyday tasks, Lupsasca found it could reproduce his best paper in 30 minutes. The team posed a year-long unsolved problem about single-minus gluon tree amplitudes; ChatGPT solved it before the professor's plane landed, discovering a "half-collinear limit" that collapsed complex results into simple formulas. When tasked with graviton research, the model produced 110 pages of novel physics in a day, leading to a verified paper on quantum gravity. Lupsasca calls this "Vibe Physics"\u2014unlike Vibe Coding, it genuinely extends the frontier of human knowledge.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n The distillation panic<\/strong>\uff08Interconnects\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Interconnects \u535a\u5ba2\u6587\u7ae0\u300a\u84b8\u998f\u6050\u614c\u300b\u6279\u8bc4\u4e86\u5c06\u90e8\u5206\u4e2d\u56fd\u5b9e\u9a8c\u5ba4\u7684 API \u6ee5\u7528\u884c\u4e3a\u79f0\u4e3a"\u84b8\u998f\u653b\u51fb"\u7684\u672f\u8bed\u4f7f\u7528\u3002\u4f5c\u8005\u6307\u51fa\uff0c\u867d\u7136\u67d0\u4e9b\u4e2d\u56fd\u5b9e\u9a8c\u5ba4\u786e\u5b9e\u901a\u8fc7\u8d8a\u72f1\u3001\u9ed1\u5ba2\u653b\u51fb\u7b49\u624b\u6bb5\u7ed5\u8fc7 API \u9650\u5236\u4ee5\u83b7\u53d6\u66f4\u591a\u8bad\u7ec3\u4fe1\u53f7\uff0c\u4f46"\u84b8\u998f\u653b\u51fb"\u8fd9\u4e00\u672f\u8bed\u4f1a\u4e0d\u53ef\u633d\u56de\u5730\u5c06\u6240\u6709\u84b8\u998f\u6280\u672f\u4e0e\u4e0d\u5f53\u884c\u4e3a\u5173\u8054\u8d77\u6765\u3002\u84b8\u998f\u672c\u8eab\u662f\u884c\u4e1a\u6807\u51c6\u7684\u540e\u8bad\u7ec3\u6280\u672f\uff0c\u88ab\u5e7f\u6cdb\u7528\u4e8e\u521b\u5efa\u66f4\u5c0f\u3001\u66f4\u4e13\u4e1a\u7684\u6a21\u578b\uff0c\u5b66\u672f\u754c\u548c\u5c0f\u578b\u4f01\u4e1a\u5c24\u5176\u4f9d\u8d56\u8fd9\u4e00\u65b9\u6cd5\u3002\u6587\u7ae0\u8b66\u544a\uff0c\u5f53\u524d\u56f4\u7ed5\u84b8\u998f\u7684\u8ba8\u8bba\u6b63\u8fc5\u901f\u6f14\u53d8\u4e3a\u76d1\u7ba1\u8fc7\u5ea6\uff0c\u5305\u62ec\u56fd\u4f1a\u59d4\u5458\u4f1a\u63a8\u8fdb\u7684\u6cd5\u6848\u3001\u767d\u5bab\u884c\u653f\u4ee4\u4ee5\u53ca\u9488\u5bf9\u4f7f\u7528\u4e2d\u56fd\u6a21\u578b\u7684\u7f8e\u56fd\u516c\u53f8\u7684\u56fd\u4f1a\u76d1\u7763\u8c03\u67e5\u3002\u8fd9\u79cd\u591a\u7ba1\u9f50\u4e0b\u7684\u76d1\u7ba1\u73af\u5883\u53ef\u80fd\u5bfc\u81f4\u4e25\u91cd\u540e\u679c\uff0c\u5982\u5b9e\u9645\u4e0a\u7981\u6b62\u4e2d\u56fd\u5f00\u53d1\u7684\u5f00\u6e90\u6743\u91cd\u6a21\u578b\uff0c\u4ece\u800c\u4f24\u5bb3\u897f\u65b9\u5b66\u672f\u754c\u548c\u5c0f\u578b\u516c\u53f8\u7684\u751f\u6001\u7cfb\u7edf\u3002\u4f5c\u8005\u5efa\u8bae\u5c06\u8fd9\u4e9b\u6ee5\u7528\u884c\u4e3a\u79f0\u4e3a"\u8d8a\u72f1"\u6216"API \u6ee5\u7528"\u800c\u975e"\u84b8\u998f\u653b\u51fb"\uff0c\u5e76\u8b66\u544a\u4ed3\u4fc3\u7684\u653f\u7b56\u53ef\u80fd\u9002\u5f97\u5176\u53cd\u2014\u2014\u5982\u679c\u5207\u65ad\u4e2d\u56fd\u516c\u53f8\u4f9d\u8d56\u7684\u84b8\u998f\u6377\u5f84\uff0c\u53cd\u800c\u53ef\u80fd\u8feb\u4f7f\u5b83\u4eec\u53d1\u5c55\u51fa\u66f4\u5177\u7ade\u4e89\u529b\u7684\u957f\u671f\u6280\u672f\u80fd\u529b\u3002<\/p>\n English Summary:<\/strong> Interconnects blog post "The Distillation Panic" critiques the term "distillation attacks" for Chinese labs' API abuse. While some labs do jailbreak and hack APIs to extract training signals, the terminology risks associating all distillation\u2014a standard industry technique for creating smaller, specialized models\u2014with illicit behavior. The author warns that discourse is snowballing into regulatory overreach, including a congressional bill, White House executive order, and oversight targeting U.S. companies using Chinese models. This multi-pronged approach could effectively ban Chinese open-weight models, harming Western academics and small companies who depend on them. The author recommends calling the abuse "jailbreaking" or "API abuse" rather than "distillation attacks," and cautions that cutting off China's distillation crutch may backfire by forcing them onto a more competitive long-term trajectory.<\/p>\n \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n Register now for OpenClaw: After Hours @ GitHub<\/strong>\uff08GitHub AI\/ML\uff09<\/p>\n \u4e2d\u6587\u6458\u8981\uff1a<\/strong>GitHub \u5ba3\u5e03\u5c06\u4e8e 2026 \u5e74 6 \u6708 3 \u65e5\u5728\u65e7\u91d1\u5c71\u603b\u90e8\u4e3e\u529e "OpenClaw: After Hours" \u6d3b\u52a8\uff0c\u4e0e Microsoft Build 2026 \u540c\u671f\u4e3e\u884c\u3002OpenClaw \u662f\u589e\u957f\u6700\u5feb\u7684\u5f00\u6e90\u9879\u76ee\u4e4b\u4e00\uff0c\u5df2\u83b7\u5f97\u8d85\u8fc7 35 \u4e07\u661f\u6807\uff0c\u62e5\u6709\u4e00\u4e2a\u79ef\u6781\u63a2\u7d22\u4ee3\u7406\u5f0f\u7cfb\u7edf\u5b9e\u9645\u80fd\u529b\u7684\u65e9\u671f\u5f00\u53d1\u8005\u793e\u533a\u3002\u6d3b\u52a8\u5c06\u5305\u62ec\u4e0e OpenClaw \u521b\u59cb\u4eba Peter Steinberger\uff08\u88ab\u79f0\u4e3a "ClawFather"\uff09\u7684\u7089\u8fb9\u5bf9\u8bdd\u3001\u4e0e\u7ef4\u62a4\u8005\u548c\u751f\u6001\u7cfb\u7edf\u6784\u5efa\u8005\u7684\u5c0f\u7ec4\u8ba8\u8bba\u3001\u5feb\u901f\u95ea\u7535\u6f14\u8bb2\u4ee5\u53ca\u8f7b\u677e\u7684\u793e\u4ea4\u65f6\u95f4\u3002\u6d3b\u52a8\u65e8\u5728\u5c06 OpenClaw \u793e\u533a\u805a\u96c6\u5230\u540c\u4e00\u7a7a\u95f4\uff0c\u8ba9\u5f00\u53d1\u8005\u4ea4\u6d41\u5b9e\u8df5\u7ecf\u9a8c\u3001\u5206\u4eab\u5728\u771f\u5b9e\u573a\u666f\u4e2d\u90e8\u7f72\u4ee3\u7406\u5f0f\u7cfb\u7edf\u7684\u7ecf\u9a8c\u4e0e\u6311\u6218\u3002\u6d3b\u52a8\u63d0\u4f9b\u73b0\u573a\u53c2\u4e0e\u548c Twitch \u76f4\u64ad\u4e24\u79cd\u65b9\u5f0f\uff0c\u5730\u70b9\u4f4d\u4e8e GitHub \u603b\u90e8\uff08275 Brannan St., San Francisco\uff09\uff0c\u65f6\u95f4\u4e3a\u665a\u4e0a 5:30 \u81f3 9 \u70b9\u3002\u7531\u4e8e\u540d\u989d\u6709\u9650\uff0c\u53c2\u4e0e\u8005\u9700\u63d0\u524d\u6ce8\u518c\u5e76\u7b49\u5f85\u786e\u8ba4\u3002<\/p>\n English Summary:<\/strong> GitHub announced "OpenClaw: After Hours" on June 3, 2026, at GitHub HQ in San Francisco during Microsoft Build 2026. OpenClaw, one of the fastest-growing open source projects with over 350,000 stars, has an early community of builders exploring what agentic systems can do in practice. The event features a fireside chat with Peter Steinberger (the "ClawFather" and OpenClaw creator), a panel with maintainers and ecosystem builders sharing what's working and not working when shipping real agentic systems, lightning talks, and a happy hour for networking. The gathering aims to bring the OpenClaw community together to trade notes on practical deployment experiences. Attendance is available in-person or via Twitch livestream at twitch.tv\/github. Spots are limited and registration requires confirmation.<\/p>\n