
{"id":9068,"date":"2026-06-13T23:46:37","date_gmt":"2026-06-13T15:46:37","guid":{"rendered":"https:\/\/infernews.com\/blog\/a-long-horizon-real-world-benchmark-for-computer-use-agents-with-hybrid-interfac\/"},"modified":"2026-06-13T23:47:38","modified_gmt":"2026-06-13T15:47:38","slug":"weavebench","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/weavebench\/","title":{"rendered":"WeaveBench\uff1a\u6e2c\u8a66 CUA \u771f\u672c\u4e8b\u7684\u57fa\u6e96"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/06\/pasted-a07e2c3910e7.jpg\" alt=\"Click to watch: an agent managing a RabbitMQ dead-letter-queue topology end-to-end\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">WeaveBench \u662f\u4e00\u500b benchmark \u57fa\u6e96\u9805\u76ee\uff0c\u805a\u7126\u6e2c\u8a66 Computer-use agents\uff08CUAs\uff09\u5728\u771f\u5be6\u684c\u9762\u74b0\u5883\u4e2d\uff0c\u80fd\u5426\u628a GUI \u9ede\u64ca\u3001shell \u6307\u4ee4\u8207\u7a0b\u5f0f\u78bc\u7de8\u4fee\u4e32\u6210\u540c\u4e00\u689d\u5de5\u4f5c\u6d41\u7a0b\u3002\u5b83\u8655\u7406\u7684\u4e0d\u662f\u55ae\u4e00\u6b65\u9a5f\u6e96\u78ba\u7387\uff0c\u800c\u662f\u9577\u6d41\u7a0b\u3001\u591a\u4ecb\u9762\u5354\u4f5c\u9019\u985e\u66f4\u63a5\u8fd1\u65e5\u5e38\u5de5\u4f5c\u7684\u554f\u984c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u9019\u500b\u9805\u76ee\u7684\u5224\u5206\u65b9\u5f0f\u6bd4\u5e38\u898b\u7684\u300c\u6709\u6c92\u6709\u751f\u6210\u67d0\u500b\u6a94\u6848\u300d\u56b4\u683c\u5f97\u591a\u3002\u5b83\u4f7f\u7528 trajectory-aware Agent-as-Judge\uff0c\u6703\u95b1\u8b80 chat trace\u3001\u4ea4\u4ed8\u7269\uff0c\u4e26\u6309\u689d\u6b3e\u63d0\u4f9b\u8b49\u64da\uff1b\u8ad6\u6587\u4ea6\u6307\u51fa\uff0c\u53ea\u770b\u6700\u7d42\u7d50\u679c\u6703\u9ad8\u4f30\u4ee3\u7406\u8868\u73fe\uff0c\u9019\u9ede\u5c0d\u7814\u7a76 CUAs \u7684\u4eba\u5f88\u6709\u53c3\u8003\u50f9\u503c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5982\u679c\u60f3\u5148\u4e86\u89e3\u5b83\u600e\u6a23\u904b\u4f5c\uff0c\u53ef\u4ee5\u5148\u770b\u96e2\u7dda demo\uff0c\u76f4\u63a5\u89c0\u5bdf score.json\u3001judge model \u56de\u61c9\u548c\u9010\u9805\u8b49\u64da\uff0c\u518d\u6c7a\u5b9a\u662f\u5426\u4e0b\u8f09\u5b8c\u6574\u8cc7\u6599\u96c6\u8207 qcow2 \u57f7\u884c\u74b0\u5883\u3002\u5b8c\u6574\u6d41\u7a0b\u9700\u8981 Linux\u3001KVM\u3001Docker \u53ca\u76f8\u7576\u591a\u8a18\u61b6\u9ad4\u8207\u78c1\u789f\u7a7a\u9593\uff0c\u8f03\u9069\u5408\u7814\u7a76\u5718\u968a\u3001\u6a21\u578b\u8a55\u6e2c\u4eba\u54e1\uff0c\u6216\u6b63\u5728\u5efa\u69cb\u4ee3\u7406\u7cfb\u7d71\u7684\u5de5\u7a0b\u5e2b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>114 \u500b\u9577\u6d41\u7a0b\u4efb\u52d9\uff0c\u6db5\u84cb 8 \u500b\u5de5\u4f5c\u9818\u57df<\/li>\n\n\n\n<li>\u6bcf\u500b\u4efb\u52d9\u90fd\u8981\u6c42 GUI \u8207 CLI\/code \u4ea4\u66ff\u64cd\u4f5c<\/li>\n\n\n\n<li>\u6700\u4f73\u516c\u958b\u7d50\u679c\u70ba 41.2% PassRate\uff0c\u986f\u793a\u96e3\u5ea6\u4ecd\u7136\u5f88\u9ad8<\/li>\n\n\n\n<li>\u63d0\u4f9b OSWorld hybrid-scoring experiment\uff0c\u53ef\u5c0d\u7167\u4e0d\u540c\u8a55\u5206\u8207\u57f7\u884c\u6846\u67b6<\/li>\n\n\n\n<li>\u8cc7\u6599\u96c6\u3001runtime \u8207 qcow2 \u5df2\u653e\u5728 \ud83e\udd17 wanlilll\/WeaveBench<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u76f8\u95dc\u6a21\u578b\u8207\u7d44\u5408\u65b9\u9762\uff0c\u516c\u958b\u7d50\u679c\u5305\u62ec Claude Opus 4.7 + Claude Code\u3001Claude Opus 4.7 + OpenClaw\u3001GPT-5.5 + Codex CLI\u3001GPT-5.5 + OpenClaw\u3001GPT-5.4 + OpenClaw\uff0c\u4ee5\u53ca Gemini 3.1 Pro + OpenClaw\u3002\u82e5\u4f60\u95dc\u5fc3\u4ee3\u7406\u662f\u5426\u771f\u7684\u61c2\u5f97\u8de8\u4ecb\u9762\u5b8c\u6210\u5de5\u4f5c\uff0c\u800c\u4e0d\u662f\u53ea\u6703\u5728\u55ae\u4e00\u6e2c\u8a66\u96c6\u5237\u5206\uff0c\u9019\u500b\u9805\u76ee\u5f88\u6709\u7814\u7a76\u50f9\u503c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>GitHub\uff1a<\/strong> <a href=\"https:\/\/github.com\/weavebench\/WeaveBench\" rel=\"noopener noreferrer\">https:\/\/github.com\/weavebench\/WeaveBench<\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u9805\u76ee\uff1a<\/strong> <a href=\"https:\/\/weavebench.github.io\/\" rel=\"noopener noreferrer\">https:\/\/weavebench.github.io\/<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>WeaveBench \u4e0d\u53ea\u662f\u984c\u5eab\uff0c\u800c\u662f\u5c08\u9580\u6aa2\u9a57 Computer-use agents \u80fd\u5426\u8de8 GUI \u8207 CLI \u5b8c\u6210\u9577\u6d41\u7a0b\u5de5\u4f5c\u7684\u57fa\u6e96\u9805\u76ee\u3002<\/p>\n","protected":false},"author":8,"featured_media":9067,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,173,196,197],"tags":[],"class_list":["post-9068","post","type-post","status-publish","format-standard","hentry","category-133","category-173","category-196","category-framework"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9068","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=9068"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9068\/revisions"}],"predecessor-version":[{"id":9070,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/9068\/revisions\/9070"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/9067"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=9068"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=9068"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=9068"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}