
{"id":11384,"date":"2026-08-25T01:02:52","date_gmt":"2026-08-24T17:02:52","guid":{"rendered":"https:\/\/infernews.com\/blog\/a-novel-benchmerk-specially-designed-to-evaluate-omni-llms-under-assistant-style\/"},"modified":"2026-08-25T01:07:32","modified_gmt":"2026-08-24T17:07:32","slug":"a-novel-benchmerk-specially-designed-to-evaluate-omni-llms-under-assistant-style","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/a-novel-benchmerk-specially-designed-to-evaluate-omni-llms-under-assistant-style\/","title":{"rendered":"OmniAssistBench\uff1a\u4ee5\u591a\u8f2a\u4e92\u52d5\u8a55\u6e2c\u5373\u6642\u5f71\u7247\u52a9\u624b\u7684\u8868\u73fe"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/pasted-d399bff89669.jpg\" alt=\"OmniAssistBench proposes an annotation pipeline which allows annotators to build test samples from existing Internet vid\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7576\u5f71\u7247\u52a9\u624b\u7684\u56de\u7b54\u6703\u6539\u8b8a\u4f7f\u7528\u8005\u4e0b\u4e00\u6b65\u884c\u52d5\uff0c\u50b3\u7d71\u53ea\u5c0d\u56fa\u5b9a\u5f71\u7247\u554f\u7b54\u7684\u6e2c\u8a66\u4fbf\u96e3\u4ee5\u53cd\u6620\u771f\u5be6\u9ad4\u9a57\u3002OmniAssistBench \u5c6c\u65bc Omni-LLMs \u7684\u5f71\u7247\u52a9\u624b\u8a55\u6e2c\u57fa\u6e96\u53ca\u6578\u64da\u96c6\uff0c\u900f\u904e\u56fa\u5b9a\u4e92\u52d5\u8def\u5f91\uff0c\u8655\u7406\u591a\u8f2a\u5c0d\u8a71\u4e2d\u5f71\u7247\u5167\u5bb9\u6703\u96a8\u6a21\u578b\u56de\u61c9\u8b8a\u5316\u7684\u554f\u984c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u9805\u76ee\u6db5\u84cb 685 \u7d44\u958b\u653e\u5f0f\u554f\u7b54\uff0c\u5206\u6210 Basic Tier \u8207 Advanced Tier\uff0c\u6d89\u53ca\u793e\u4ea4\u7406\u89e3\u3001\u6642\u9593\u611f\u77e5\u3001\u624b\u52e2\u63d0\u793a\u3001\u7a0b\u5e8f\u6307\u5c0e\u53ca\u4e3b\u52d5\u56de\u61c9\u7b49 7 \u985e\u4e3b\u8981\u4efb\u52d9\u548c 16 \u985e\u7d30\u5206\u4efb\u52d9\u3002\u5f71\u7247\u53d6\u6750\u81ea\u904b\u52d5\u3001\u716e\u98df\u3001\u8b1b\u8ab2\u3001DIY \u53ca\u8ac7\u8a71\u7bc0\u76ee\uff0c\u53e6\u6709 3 \u500b\u81ea\u884c\u62cd\u651d\u6848\u4f8b\uff0c\u6bcf\u500b\u5e73\u5747\u8d85\u904e 15 \u8f2a\u4e92\u52d5\uff1b\u554f\u984c\u6703\u4ee5\u81ea\u7136\u8a9e\u97f3\u3001\u6253\u5b57\u6216\u624b\u5beb\u65b9\u5f0f\u5d4c\u5165\u5f71\u7247\uff0c\u800c\u4e0d\u662f\u53ea\u653e\u6587\u5b57\u63d0\u793a\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u5f71\u7247\u9023\u540c\u5c0d\u61c9\u97f3\u8a0a\u8f38\u5165\u6a21\u578b\uff0c\u518d\u6309\u4e92\u52d5\u56de\u5408\u56de\u7b54\u554f\u984c<\/li>\n\n\n\n<li>\u7531\u5c08\u5bb6\u5148\u6574\u7406\u5f71\u7247\u6b65\u9a5f\uff0c\u518d\u526a\u8f2f\u7247\u6bb5\u53ca\u52a0\u5165\u5b57\u5e55\uff0c\u56fa\u5b9a\u6e2c\u8a66\u8def\u5f91<\/li>\n\n\n\n<li>\u8a55\u5206\u63a1\u7528 LLM-as-a-Judge\uff0c\u539f\u59cb 0 \u81f3 5 \u5206\u518d\u63db\u7b97\u70ba 0 \u81f3 100<\/li>\n\n\n\n<li>\u5efa\u7acb\u6578\u64da\u96c6\u8017\u7528\u8d85\u904e 1,000 \u5c0f\u6642\u5c08\u5bb6\u5de5\u6642<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u9019\u7a2e\u505a\u6cd5\u6bd4\u975c\u614b\u5f71\u7247\u554f\u7b54\u66f4\u8cbc\u8fd1\u52a9\u624b\u5834\u666f\uff0c\u4f46\u56fa\u5b9a\u8def\u5f91\u4ea6\u4ee3\u8868\u5176\u4ed6\u5408\u7406\u56de\u61c9\u672a\u5fc5\u80fd\u5728\u540c\u4e00\u6e2c\u8a66\u4e2d\u7372\u5f97\u5145\u5206\u53cd\u6620\u3002Project homepage \u7684\u6392\u884c\u699c\u986f\u793a\uff0cGemini-3-Pro \u6574\u9ad4\u5f97\u5206\u70ba 66.4\uff0cGemini-2.5-Pro \u70ba 64.6\uff0c\u5546\u7528\u53ca\u958b\u6e90 Omni-LLMs \u90fd\u53ea\u9054\u90e8\u5206\u6b63\u78ba\uff0c\u53cd\u6620\u53ef\u9760\u7684\u5373\u6642\u5f71\u7247\u52a9\u624b\u4ecd\u6709\u660e\u986f\u6539\u9032\u7a7a\u9593\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u7814\u7a76\u5718\u968a\u4f86\u81ea\u5357\u4eac\u5927\u5b78\u3001\u5357\u958b\u5927\u5b78\u53ca University of Waterloo\u3002\u6e2c\u8a66\u8005\u9700\u8981\u6309\u7167\u9805\u76ee\u63d0\u4f9b\u7684\u8a55\u6e2c\u6d41\u7a0b\u8655\u7406\u5f71\u7247\u548c\u97f3\u8a0a\uff0c\u76f8\u95dc Dataset\u3001Project Page \u53ca\u8a55\u4f30\u7a0b\u5f0f\u78bc\u53ef\u4f9b\u67e5\u95b1\uff1b\u6b63\u5f0f\u63a1\u7528\u524d\u61c9\u5148\u78ba\u8a8d\u7248\u672c\u3001\u5b58\u53d6\u6b0a\u9650\u53ca\u8a55\u5206\u6d41\u7a0b\u662f\u5426\u5df2\u66f4\u65b0\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/xianyunsun.github.io\/OmniAssistBench\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/XianyunSun\/OmniAssistBench\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>OmniAssistBench \u4ee5 LLM-as-a-Judge \u56fa\u5b9a\u4e92\u52d5\u8def\u5f91\u6e2c\u8a66 Omni-LLMs \u7684\u591a\u8f2a\u5f71\u7247\u52a9\u624b\u80fd\u529b\uff0c\u6db5\u84cb685\u7d44\u958b\u653e\u5f0f\u554f\u7b54\u30017\u985e\u4efb\u52d9\uff0c\u5f71\u7247\u542b\u81ea\u7136\u8a9e\u97f3\u3001\u6253\u5b57\u6216\u624b\u5beb\u63d0\u793a\u3002\u8a55\u5206\u986f\u793a Gemini-3-Pro \u4ee566.4\u5206\u9818\u5148\u3002<\/p>\n","protected":false},"author":8,"featured_media":11383,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wpai_generated_summary":"","footnotes":""},"categories":[133,185,176,140,198,199],"tags":[],"class_list":["post-11384","post","type-post","status-publish","format-standard","hentry","category-133","category-qwen","category-176","category-gemini","category-198","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11384","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=11384"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11384\/revisions"}],"predecessor-version":[{"id":11387,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11384\/revisions\/11387"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/11383"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=11384"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=11384"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=11384"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}