{"id":9730,"date":"2026-11-21T09:56:08","date_gmt":"2026-11-21T09:56:08","guid":{"rendered":"https:\/\/i10xnews.kinsta.cloud\/?p=9730"},"modified":"2026-11-21T08:15:18","modified_gmt":"2026-11-21T08:15:18","slug":"gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","status":"publish","type":"post","link":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","title":{"rendered":"GPT-6.2 Astra Tops Blueprint-Bench 2, 70% Above May Leader GPT-5.5"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">OpenAI&#8217;s GPT-6.2 Astra now ranks first on Blueprint-Bench 2, Andon Labs&#8217; test of whether AI agents can turn apartment photos into correct floor plans, with a score of 0.614 on the <a href=\"https:\/\/andonlabs.com\/evals\/blueprint-bench-2\" target=\"_blank\" rel=\"noreferrer noopener\">public leaderboard<\/a>. That is a hair above the 0.586 human baseline and 70% above GPT-5.5, which led the board when the test launched in May. Andon frames the result more cautiously than the headlines: it says GPT-6.2 Astra and Gemini 4 Carbon now tie the humans, and its <a href=\"https:\/\/x.com\/andonlabs\/status\/2123570732575989832\" target=\"_blank\" rel=\"noreferrer noopener\">announcement on X<\/a> calls Astra the AI that understands the physical 3D world best.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">How the test works<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Each agent works through 50 apartments in a row and gets about 20 interior photos per apartment. From those it has to draw a 2D plan that shows the rooms, which rooms connect through doors and their rough sizes. A persistent notepad lets the agent carry lessons from one apartment to the next.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Andon scores each plan by comparing its connection graph with the real layout. Half the score comes from the overlap of room-to-room connections, the rest from how many doors each room has, overall connectivity, room count, door count and orientation. Scores are normalized so that random guessing equals 0 and a perfect plan equals 1.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">The numbers behind the headline<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Our reading of the leaderboard shows a tight race at the top and some odd results further down:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>Five models within 0.07:<\/strong> GPT-6.2 Astra 0.614, Gemini 4 Carbon 0.586, GPT-6.2 Sol 0.570, Claude Fable 5.2 0.566 and Gemini 4 Argon 0.544.<\/li><li><strong>OpenAI&#8217;s jump:<\/strong> GPT-6 Astra scored 0.497, so GPT-6.2 Astra adds 0.117 points, about 24%. On the Sol line the gain is larger, from 0.432 for GPT-6.1 Sol to 0.570, about 32%.<\/li><li><strong>Anthropic&#8217;s step:<\/strong> Claude Fable 5.2 improves on Fable 5.1 (0.419) by about 35%, while Claude Opus 4.8 sits at 0.145, below the older Opus 4.7 at 0.245.<\/li><li><strong>Grok goes backwards:<\/strong> Grok 4.7 (0.325) lands just below Grok 4.6 (0.332).<\/li><\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Two caveats matter before anyone writes &#8220;AI beats humans&#8221;. The human baseline was measured on only 12 of the 50 apartments, and the 0.028 lead is small next to the gaps between model versions. Andon shows standard errors in its chart but does not list them in the table.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Six months of progress, and where models still fail<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">When Blueprint-Bench 2 went live in May, Andon named GPT-5.5, Gemini 3.1 Pro and Claude Opus 4.7 as the leading trio. On today&#8217;s board they score 0.362, 0.265 and 0.245. The new leader has added 0.252 points in roughly six months, which matches the trend of about 0.04 points per month that Andon fits through frontier releases. If that line simply continued, the top score would pass 0.8 in under five months. Trend lines on young benchmarks rarely stay that straight, so we read this as a measure of momentum, not a forecast.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The weak spot is also clear from Andon&#8217;s notes. According to Andon&#8217;s launch notes, models get the room count right about nine times out of ten. What separates the field is which room connects to which, the part that carries half the score. In practice a model can tell you an apartment has a kitchen, two bedrooms and a hallway, but it still often misplaces the doors between them.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Specialist models are missing the newest version<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Gemini Robotics-ER 1.6, built for spatial and embodied reasoning, scores at or below random. That looks like a verdict on specialist robot models, but it is an older one. Our <a href=\"https:\/\/i10x.ai\/news\/engineering-robotics\/robot-models\">Robot AI Model Comparison<\/a> lists Gemini Robotics 2 from July 2026, which ships with a new ER 2 reasoning model. Andon has not tested ER 2, so the board says little about Google&#8217;s current robotics stack.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The result does fit a bet OpenAI made last week. The company <a href=\"https:\/\/i10x.ai\/news\/openai-disbands-robotics-team-general-models-gpt-6-astra\">folded its stand-alone robotics team into its main research group<\/a>, arguing that general models like GPT-6 Astra can be applied to robots directly. A general model topping a spatial test is one piece of evidence for that view. Andon&#8217;s own Butter-Bench, where LLMs steer a real office robot, is the reminder that reading a room from photos and moving through it are different skills: among the models tested there, the best completed 40% of tasks against 95% for humans.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Which model to pick for spatial work<\/h2>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>Price matters at the top:<\/strong> GPT-6.2 Astra lists at $8 per million input and $40 per million output tokens, as we reported at <a href=\"https:\/\/i10x.ai\/news\/openai-gpt-6-2-astra-sol-launch\">its launch<\/a>. <a href=\"https:\/\/i10x.ai\/news\/gemini-4-carbon-now-available-api-ultra-pricing\">Gemini 4 Carbon<\/a> costs $2 and $10 until 30 November, then $4 and $20. For a 0.028 difference on this test, Carbon is a quarter of the price this month and half of it afterwards.<\/li><li><strong>Look beyond one board:<\/strong> on Andon&#8217;s <a href=\"https:\/\/andonlabs.com\/evals\/vending-bench-2\" target=\"_blank\" rel=\"noreferrer noopener\">Vending-Bench 2<\/a>, a year-long business simulation, Gemini 4 Neon and Gemini 4 Carbon lead and GPT-6 Astra is third. GPT-6.2 has not been listed there yet.<\/li><li><strong>For real estate, AR and home-scanning apps:<\/strong> treat model-drawn plans as a draft that a person checks, especially door positions. Room counts are reliable, connections are not yet.<\/li><\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2d i10x Perspective<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">In Andon&#8217;s original Blueprint-Bench, model output was essentially noise. On the second version the best models now match the human baseline. That shifts the bottleneck from the model&#8217;s understanding of a room to the hardware and control layer that has to act on it. We expect the next round of robot demos to lean on general frontier models for planning, exactly as OpenAI now argues.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The leaderboard is also a reminder that a single number is a poor buying guide. Astra wins here by 0.028 against a human comparison that covers only 12 apartments, Carbon leads on long-horizon business tasks, and Carbon costs far less. Pick by task and price, not by who is first this week.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Still open<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Andon has not said which reasoning settings it used for the GPT-6.2 runs, and there is no new human baseline across all 50 apartments. Gemini Robotics ER 2 and GPT-6.2 on Vending-Bench 2 are the two obvious missing entries. Whether the lead holds once the next Gemini and Claude releases are tested is the next check; we will update this article when the board changes.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above the May leader. Our read of six months of progress, where models still fail and how price changes the pick.<\/p>\n","protected":false},"author":2,"featured_media":9729,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_yoast_wpseo_meta-robots-noindex":"","paper_title":"","paper_url":"","pdf_url":"","doi":"","arxiv_id":"","authors":"","institutions":"","paper_date":"","venue":"","peer_reviewed":false,"code_url":"","data_url":"","funding_coi":"","figure_license":"","paper_license":"","replication":"","study_models":"","study_sample":"","nx_approved":"1","footnotes":"","_yoast_wpseo_primary_category":"9","_nx_sec_severity":"","_nx_sec_affected":"","_nx_sec_status":"","_nx_sec_kicker":"","_nx_sec_type":"","_nx_sec_vendor":"","_nx_sec_fixed":"","_nx_sec_action":""},"categories":[9],"tags":[],"companies":[38],"products":[],"class_list":["post-9730","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-models","company-open-ai"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News<\/title>\n<meta name=\"description\" content=\"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May&#039;s leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News\" \/>\n<meta property=\"og:description\" content=\"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May&#039;s leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\" \/>\n<meta property=\"og:site_name\" content=\"i10X News\" \/>\n<meta property=\"article:published_time\" content=\"2026-11-21T09:56:08+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1344\" \/>\n\t<meta property=\"og:image:height\" content=\"768\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"Christopher Ort\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:image\" content=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/i10x-news-og-default-1200x630-1.png\" \/>\n<meta name=\"twitter:creator\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:site\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Christopher Ort\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\"},\"author\":{\"name\":\"Christopher Ort\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\"},\"headline\":\"GPT-6.2 Astra Tops Blueprint-Bench 2, 70% Above May Leader GPT-5.5\",\"datePublished\":\"2026-11-21T09:56:08+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\"},\"wordCount\":966,\"publisher\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/11\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp\",\"articleSection\":[\"Models\"],\"inLanguage\":\"en-US\",\"dateModified\":\"2026-11-21T09:56:08+00:00\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\",\"name\":\"Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/11\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp\",\"datePublished\":\"2026-11-21T09:56:08+00:00\",\"description\":\"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May's leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/11\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/11\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp\",\"width\":1344,\"height\":768,\"caption\":\"Image: AI-generated, i10X\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/i10x.ai\\\/news\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"GPT-6.2 Astra Tops Blueprint-Bench 2, 70% Above May Leader GPT-5.5\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#website\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\",\"name\":\"i10X News\",\"description\":\"Stay updated with the latest news and announcements from i10X.\",\"publisher\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/i10x.ai\\\/news?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\",\"name\":\"i10X\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"contentUrl\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"width\":180,\"height\":180,\"caption\":\"i10X\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/x.com\\\/i10x_ai\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/i10x-ai\",\"https:\\\/\\\/instagram.com\\\/i10x.ai\",\"https:\\\/\\\/www.youtube.com\\\/@i10X_ai_official\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\",\"name\":\"Christopher Ort\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"caption\":\"Christopher Ort\"},\"description\":\"Christopher Ort is the editor of i10X News at i10x.ai. He has worked hands-on with AI tools, models and agent workflows for years and leads i10X\u2019s coverage of AI launches, funding and research. He is the author of \u201cThe Style Penalty\u201d, an SSRN study on how frontier AI models judge AI-written CVs, covered by heise online and t3n.\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/author\\\/christopher-ort\",\"sameAs\":[\"https:\\\/\\\/i10x.ai\\\/blog\\\/author\\\/christopher-ort\"],\"worksFor\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\"},\"alumniOf\":{\"@type\":\"EducationalOrganization\",\"name\":\"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\",\"url\":\"https:\\\/\\\/www.fau.eu\"},\"subjectOf\":[{\"@type\":\"CreativeWork\",\"url\":\"https:\\\/\\\/papers.ssrn.com\\\/abstract=7426699\",\"name\":\"The Style Penalty: How AI Resume-Writing Tools Shape Outcomes in Automated Hiring Screens\"},{\"@type\":\"CreativeWork\",\"url\":\"https:\\\/\\\/www.heise.de\\\/news\\\/KI-Chatbots-bewerten-KI-Lebenslaeufe-Claude-findet-sich-besser-als-GPT-11317531.html\",\"name\":\"KI-Chatbots bewerten KI-Lebensl\u00e4ufe: Claude findet sich besser als GPT\"},{\"@type\":\"CreativeWork\",\"url\":\"https:\\\/\\\/t3n.de\\\/news\\\/bewerbung-schreiben-ki-5-fehler-1747190\\\/\",\"name\":\"Bewerbung mit KI schreiben: Diese 5 Fehler solltet ihr laut Studien vermeiden\"}]}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News","description":"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May's leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","og_locale":"en_US","og_type":"article","og_title":"Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News","og_description":"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May's leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.","og_url":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","og_site_name":"i10X News","article_published_time":"2026-11-21T09:56:08+00:00","og_image":[{"width":1344,"height":768,"url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp","type":"image\/webp"}],"author":"Christopher Ort","twitter_card":"summary_large_image","twitter_image":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/i10x-news-og-default-1200x630-1.png","twitter_creator":"@i10x_ai","twitter_site":"@i10x_ai","twitter_misc":{"Written by":"Christopher Ort","Est. reading time":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#article","isPartOf":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline"},"author":{"name":"Christopher Ort","@id":"https:\/\/i10x.ai\/news\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a"},"headline":"GPT-6.2 Astra Tops Blueprint-Bench 2, 70% Above May Leader GPT-5.5","datePublished":"2026-11-21T09:56:08+00:00","mainEntityOfPage":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline"},"wordCount":966,"publisher":{"@id":"https:\/\/i10x.ai\/news\/#organization"},"image":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp","articleSection":["Models"],"inLanguage":"en-US","dateModified":"2026-11-21T09:56:08+00:00"},{"@type":"WebPage","@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","url":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline","name":"Blueprint-Bench 2: GPT-6.2 Astra Leads the Floor Plan Test - i10X News","isPartOf":{"@id":"https:\/\/i10x.ai\/news\/#website"},"primaryImageOfPage":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage"},"image":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp","datePublished":"2026-11-21T09:56:08+00:00","description":"GPT-6.2 Astra leads Blueprint-Bench 2 at 0.614, 70% above May's leader. Our read: six months of progress, where models still fail, price vs Gemini 4 Carbon.","breadcrumb":{"@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#primaryimage","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/11\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline-title.webp","width":1344,"height":768,"caption":"Image: AI-generated, i10X"},{"@type":"BreadcrumbList","@id":"https:\/\/i10x.ai\/news\/gpt-6-2-astra-blueprint-bench-2-floor-plans-human-baseline#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/i10x.ai\/news"},{"@type":"ListItem","position":2,"name":"GPT-6.2 Astra Tops Blueprint-Bench 2, 70% Above May Leader GPT-5.5"}]},{"@type":"WebSite","@id":"https:\/\/i10x.ai\/news\/#website","url":"https:\/\/i10x.ai\/news","name":"i10X News","description":"Stay updated with the latest news and announcements from i10X.","publisher":{"@id":"https:\/\/i10x.ai\/news\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/i10x.ai\/news?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/i10x.ai\/news\/#organization","name":"i10X","url":"https:\/\/i10x.ai\/news\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/#\/schema\/logo\/image\/","url":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","contentUrl":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","width":180,"height":180,"caption":"i10X"},"image":{"@id":"https:\/\/i10x.ai\/news\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/i10x_ai","https:\/\/www.linkedin.com\/company\/i10x-ai","https:\/\/instagram.com\/i10x.ai","https:\/\/www.youtube.com\/@i10X_ai_official"]},{"@type":"Person","@id":"https:\/\/i10x.ai\/news\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a","name":"Christopher Ort","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","caption":"Christopher Ort"},"description":"Christopher Ort is the editor of i10X News at i10x.ai. He has worked hands-on with AI tools, models and agent workflows for years and leads i10X\u2019s coverage of AI launches, funding and research. He is the author of \u201cThe Style Penalty\u201d, an SSRN study on how frontier AI models judge AI-written CVs, covered by heise online and t3n.","url":"https:\/\/i10x.ai\/news\/author\/christopher-ort","sameAs":["https:\/\/i10x.ai\/blog\/author\/christopher-ort"],"worksFor":{"@id":"https:\/\/i10x.ai\/news\/#organization"},"alumniOf":{"@type":"EducationalOrganization","name":"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)","url":"https:\/\/www.fau.eu"},"subjectOf":[{"@type":"CreativeWork","url":"https:\/\/papers.ssrn.com\/abstract=7426699","name":"The Style Penalty: How AI Resume-Writing Tools Shape Outcomes in Automated Hiring Screens"},{"@type":"CreativeWork","url":"https:\/\/www.heise.de\/news\/KI-Chatbots-bewerten-KI-Lebenslaeufe-Claude-findet-sich-besser-als-GPT-11317531.html","name":"KI-Chatbots bewerten KI-Lebensl\u00e4ufe: Claude findet sich besser als GPT"},{"@type":"CreativeWork","url":"https:\/\/t3n.de\/news\/bewerbung-schreiben-ki-5-fehler-1747190\/","name":"Bewerbung mit KI schreiben: Diese 5 Fehler solltet ihr laut Studien vermeiden"}]}]}},"_links":{"self":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/9730","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/comments?post=9730"}],"version-history":[{"count":4,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/9730\/revisions"}],"predecessor-version":[{"id":9741,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/9730\/revisions\/9741"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media\/9729"}],"wp:attachment":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media?parent=9730"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/categories?post=9730"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/tags?post=9730"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/companies?post=9730"},{"taxonomy":"product","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/products?post=9730"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}