{"id":11547,"date":"2026-08-27T17:16:10","date_gmt":"2026-08-27T09:16:10","guid":{"rendered":"https:\/\/infernews.com\/blog\/vgi-bench-probing-visual-intelligence-in-video-generative-models\/"},"modified":"2026-08-27T17:19:07","modified_gmt":"2026-08-27T09:19:07","slug":"vgi-bench-probing-visual-intelligence-in-video-generative-models","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/vgi-bench-probing-visual-intelligence-in-video-generative-models\/","title":{"rendered":"VGI-Bench \u63ed\u793a\u5f71\u7247\u6a21\u578b\u63a8\u7406\u4ecd\u672a\u53ef\u9760"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/pasted-cb56725b6274.jpg\" alt=\"Repository image for hexuan21\/VGI-Bench\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7576\u5f71\u7247\u751f\u6210\u6a21\u578b\u8981\u8655\u7406\u7a7a\u9593\u95dc\u4fc2\u3001\u6642\u9593\u8b8a\u5316\u3001\u7269\u7406\u64cd\u4f5c\u6216\u7d50\u69cb\u8b0e\u984c\u6642\uff0c\u756b\u9762\u597d\u770b\u4e26\u4e0d\u4ee3\u8868\u63a8\u7406\u6b63\u78ba\u3002VGI-Bench\u5c6c\u65bc\u5f71\u7247\u6a21\u578b\u8996\u89ba\u63a8\u7406\u8a55\u6e2c\u57fa\u6e96\uff0c\u5be6\u969b\u8655\u7406\u7684\u662f\u5982\u4f55\u6aa2\u67e5\u6a21\u578b\u80fd\u5426\u751f\u6210\u7b26\u5408\u689d\u4ef6\u3001\u800c\u4e14\u904e\u7a0b\u9023\u8cab\u7684\u5f71\u7247\uff0c\u800c\u975e\u53ea\u4ea4\u51fa\u4e00\u500b\u5408\u7406\u7684\u6700\u5f8c\u756b\u9762\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">VGI-Bench\u5305\u542b27\u9805\u4efb\u52d9\u53ca810\u500b\u6e2c\u8a66\u5be6\u4f8b\uff0c\u6309\u4efb\u52d9\u9818\u57df\u548c skill tags \u5206\u6210\u5169\u5c64\u5206\u985e\uff0c\u4e26\u8a2d\u6709 easy\u3001mid\u3001hard \u4e09\u500b\u96e3\u5ea6\u7d1a\u5225\u3002\u6e2c\u8a66\u8f38\u5165\u76e1\u91cf\u8cbc\u8fd1\u73fe\u6642\u5f71\u7247\u6a21\u578b\u719f\u6089\u7684\u8996\u89ba\u5148\u9a57\uff0c\u540c\u6642\u8981\u6c42\u8f38\u51fa\u53cd\u6620\u6709\u6548\u7684 evolving process\uff1b\u8a55\u5206\u63a1\u7528 rubric score \u00d7 completeness score\uff0c\u517c\u9867\u7b54\u6848\u662f\u5426\u6b63\u78ba\u53ca\u662f\u5426\u5b8c\u6210\u8981\u6c42\u3002<\/p>\n\n\n<div\n    \tclass=\"align wp-block-vpb-video\"    id='vpbpVideoPlayer-1'\n    data-attributes='{&quot;source&quot;:&quot;https:\\\/\\\/huggingface.co\\\/datasets\\\/hexuan21\\\/VGI-Bench\\\/resolve\\\/main\\\/eval_res\\\/_preview\\\/video_model_res\\\/mnx_h3\\\/remove_obstacle\\\/remove_obstacle_lv1_002.mp4&quot;,&quot;repeat&quot;:true,&quot;autoplay&quot;:true,&quot;muted&quot;:true,&quot;align&quot;:&quot;&quot;,&quot;poster&quot;:&quot;&quot;,&quot;controls&quot;:{&quot;play-large&quot;:true,&quot;restart&quot;:false,&quot;rewind&quot;:true,&quot;play&quot;:true,&quot;fast-forward&quot;:true,&quot;progress&quot;:true,&quot;current-time&quot;:true,&quot;duration&quot;:false,&quot;mute&quot;:true,&quot;volume&quot;:true,&quot;pip&quot;:false,&quot;airplay&quot;:false,&quot;captions&quot;:false,&quot;settings&quot;:true,&quot;download&quot;:false,&quot;fullscreen&quot;:true},&quot;captions&quot;:[],&quot;ratio&quot;:&quot;16:9&quot;,&quot;accentColor&quot;:&quot;&quot;,&quot;lazyLoad&quot;:false,&quot;width&quot;:&quot;100%&quot;,&quot;radius&quot;:&quot;0px&quot;,&quot;resetOnEnd&quot;:false,&quot;autoHideControl&quot;:true,&quot;isSetup&quot;:false}'\n><\/div>\n\n\n<p class=\"wp-block-paragraph\">\u7d50\u679c\u53cd\u6620\u6a21\u578b\u80fd\u529b\u4ecd\u6709\u660e\u986f\u7f3a\u53e3\uff1aSeedance 2.0\u4ee551.0%\u53d6\u5f97\u6700\u9ad8\u7e3d\u5206\uff0c\u5728 Visual Organization\u3001Spatiotemporal \u548c Physical Manipulation \u9818\u57df\u9818\u5148\uff1bMiniMax-H3\u5247\u4ee550.6%\u5728 Structured Puzzles \u9818\u5148\u3002Sora 2\u3001Gen-4.5\u3001Wan 2.7\u53caVeo 3.1\u7b49\u6a21\u578b\u4ea6\u88ab\u7d0d\u5165\u6bd4\u8f03\uff0c\u65b9\u4fbf\u5206\u8fa8\u5f71\u7247\u54c1\u8cea\u8207\u8996\u89ba\u63a8\u7406\u80fd\u529b\u4e4b\u9593\u7684\u5dee\u8ddd\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u751f\u6210\u756b\u9762\u5408\u7406\uff0c\u4e0d\u7b49\u65bc\u63a8\u7406\u904e\u7a0b\u6b63\u78ba<\/li>\n\n\n\n<li>\u8986\u84cb\u8996\u89ba\u7d44\u7e54\u3001\u6642\u7a7a\u63a8\u7406\u3001\u7d50\u69cb\u8b0e\u984c\u53ca\u7269\u7406\u64cd\u4f5c<\/li>\n\n\n\n<li>\u6700\u9ad8\u7e3d\u5206\u53ea\u670951.0%\uff0c\u53ef\u9760\u6027\u4ecd\u4e0d\u8db3<\/li>\n\n\n\n<li>\u5206\u6790\u6db5\u84cb\u8f38\u51fa\u5931\u6557\u6a21\u5f0f\u53ca\u8f38\u5165\u689d\u4ef6\u654f\u611f\u5ea6<\/li>\n\n\n\n<li>\u53bb\u566a\u5f8c\u671f\u4e3b\u8981\u4fee\u6574\u65e9\u671f\u5047\u8a2d\uff0c\u672a\u5fc5\u80fd\u4fee\u6b63\u63a8\u7406\u932f\u8aa4<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u7814\u7a76\u5718\u968a\u4ea6\u6aa2\u67e5 synthetic fine-tuning \u7684\u6027\u80fd\u8f49\u79fb\u908a\u754c\uff0c\u4e26\u5f9e internal denoising \u89d2\u5ea6\u5206\u6790\u6a21\u578b\u5982\u4f55\u4fee\u6b63\u7b54\u6848\u3002\u7d50\u679c\u6307\u5411\u6709\u9650\u7684 self-correction\uff1a\u5f8c\u7e8c\u6b65\u9a5f\u591a\u6578\u662f\u5728\u5b8c\u5584\u65e9\u671f\u5047\u8a2d\uff0c\u800c\u4e0d\u662f\u91cd\u65b0\u63a8\u7ffb\u932f\u8aa4\u63a8\u7406\u3002\u9019\u500b\u57fa\u6e96\u8f03\u9069\u5408\u5f71\u7247\u6a21\u578b\u7814\u7a76\u5718\u968a\u3001\u6a21\u578b\u4f9b\u61c9\u5546\u53ca\u9700\u8981\u6bd4\u8f03\u751f\u6210\u5f0f\u8996\u89ba\u63a8\u7406\u80fd\u529b\u7684\u8a55\u6e2c\u9805\u76ee\uff1bGitHub\u8cc7\u6599\u63d0\u4f9b project page \u53ca Hugging Face Data &amp; Res \u9023\u7d50\uff0c\u4f46\u6240\u7d66\u8cc7\u6599\u6c92\u6709\u5217\u51fa\u5b89\u88dd\u6b65\u9a5f\u6216\u5b8c\u6574\u6e2c\u8a66\u6d41\u7a0b\uff0c\u4e0d\u80fd\u76f4\u63a5\u5047\u5b9a\u53ef\u4e0b\u8f09\u5f8c\u5373\u6642\u91cd\u73fe\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/hexuan21.github.io\/VGI-Bench\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/hexuan21\/VGI-Bench\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/huggingface.co\/datasets\/hexuan21\/VGI-Bench\" rel=\"noopener noreferrer\"><strong>\u6578\u64da\u96c6<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u5f71\u7247\u751f\u6210\u6a21\u578b\u4e0d\u53ea\u8981\u756b\u9762\u5408\u7406\uff0c\u9084\u8981\u4ee4\u4e8b\u4ef6\u6309\u6b63\u78ba\u904e\u7a0b\u6f14\u8b8a\u3002VGI-Bench\u4ee527\u9805\u4efb\u52d9\u6e2c\u8a66\u9019\u7a2e\u80fd\u529b\uff0c\u6700\u9ad8\u5206\u53ea\u670951%\u3002<\/p>\n","protected":false},"author":8,"featured_media":11546,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wpai_generated_summary":"","footnotes":""},"categories":[133,127,184,199,201,203],"tags":[],"class_list":["post-11547","post","type-post","status-publish","format-standard","hentry","category-133","category-127","category-robotic","category-dataset-","category-skill-","category-minimax"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11547","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=11547"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11547\/revisions"}],"predecessor-version":[{"id":11548,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11547\/revisions\/11548"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/11546"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=11547"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=11547"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=11547"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}