{"id":6144,"date":"2026-10-19T16:49:23","date_gmt":"2026-10-19T16:49:23","guid":{"rendered":"https:\/\/i10xnews.kinsta.cloud\/?p=6144"},"modified":"2026-10-19T17:21:00","modified_gmt":"2026-10-19T17:21:00","slug":"random-attention-kv-cache-eviction-reasoning","status":"publish","type":"post","link":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning","title":{"rendered":"Randomly Evicting KV Cache Matches Smart Selectors \u2014 and Runs 32\u201343% Faster"},"content":{"rendered":"\n<h2 class=\"wp-block-heading\">\u26a1 Quick Take<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>The finding:<\/strong> Across four models and six reasoning tasks, randomly evicting within each head \u2014 while always keeping the prompt \u2014 matches the strongest scored selector on accuracy.<\/li><li><strong>Why random works:<\/strong> The prompt is the fragile part; the reasoning trace protects itself via textual restatement and cross-head redundancy.<\/li><li><strong>The speedup:<\/strong> With vLLM, Random Attention delivers 32\u201343% higher tokens\/sec at 32k-token generations because it never runs a scoring pass.<\/li><\/ul>\n\n\n<p class=\"wp-block-paragraph\">Long chains of thought blow up the KV cache. Almost every compression method scores tokens and keeps the top ones. Salesforce AI Research and UIUC collaborators show the scoring signal contributes almost nothing \u2014 if you simply refuse to evict the prompt.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1344\" height=\"768\" src=\"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-throughput.webp\" alt=\"Stat card: accuracy matches strongest baseline, 32 to 43 percent higher throughput, 4 models and 6 tasks.\" class=\"wp-image-6143\" srcset=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-throughput.webp 1344w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-throughput-300x171.webp 300w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-throughput-1024x585.webp 1024w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-throughput-768x439.webp 768w\" sizes=\"auto, (max-width: 1344px) 100vw, 1344px\" \/><figcaption class=\"wp-element-caption\">i10X chart based on data from Wang et al. (2026), arXiv:2609.03430, abstract and Figure 1b (CC BY 4.0).<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2c What the study shows<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Main grid at ~4\u00d7 compression (~3\u00d7 for LiveCodeBench) on Qwen3-4B\/14B\/32B and Phi-4-reasoning. Significance gated by paired problem-clustered bootstrap CIs and sign tests.<\/p>\n\n\n<ul class=\"wp-block-list\"><li><strong>No selector significantly beats random:<\/strong> Apparent leads (e.g., TriAttention +0.3\u20130.6 on some GPQA cells) sit inside noise.<\/li><li><strong>Scored methods can lose:<\/strong> SnapKV trails significantly on every model; R-KV and VaSE also lose on some models.<\/li><li><strong>Prompt preservation explains most gaps:<\/strong> Once the prompt is protected, selection heuristics add little.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u2705 Why it matters: complexity that does not pay<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Production reasoning stacks pay latency for elaborate eviction heuristics. Random Attention suggests most of that complexity is optional.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83e\uddea Method in brief<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Uniform random eviction within heads after protecting prompt tokens; vLLM integration; clustered bootstrap testing.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u2753 What remains open<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Reasoning traces:<\/strong> Designed for CoT redundancy; other workloads may differ.<\/li><li><strong>Compression ratio:<\/strong> Main results near 4\u00d7.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\ud83e\udded Context<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Related KV efficiency in <a href=\"\/news\/category\/studies\/efficiency-training\">Efficiency &amp; Training<\/a> and <a href=\"\/news\/category\/studies\/reasoning\">Reasoning<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udee0\ufe0f What it means for serving stacks<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Never drop the prompt:<\/strong> That is the fragile region.<\/li><li><strong>Try random eviction:<\/strong> Benchmark before fancy scorers.<\/li><li><strong>Measure tokens\/sec:<\/strong> Accuracy parity is not enough.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u270d\ufe0f About the analysis<\/h2>\n\n\n<p class=\"wp-block-paragraph\">This article is based on the full text of the preprint &#8220;Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning&#8221; (arXiv:2609.03430v2, submitted September 3, 2026). All numbers are taken from the paper&#8217;s text, tables and figures; i10X did not rerun the experiments. The chart was redrawn by i10X from abstract and Figure 1b of the paper, which is published under a CC BY 4.0 license.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2d i10x Perspective<\/h2>\n\n\n<p class=\"wp-block-paragraph\">The best eviction policy being &#8220;keep the prompt, shuffle the rest&#8221; is a deliciously deflationary result. It will annoy a lot of KV-compression papers \u2014 and save a lot of GPUs.<\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>Salesforce&#8217;s Random Attention keeps the prompt and randomly drops reasoning-trace tokens per head. Across 4 models and 6 tasks it matches the best scorer while boosting vLLM throughput 32\u201343%.<\/p>\n","protected":false},"author":2,"featured_media":6294,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"paper_title":"Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning","paper_url":"https:\/\/arxiv.org\/abs\/2609.03430","pdf_url":"https:\/\/arxiv.org\/pdf\/2609.03430v2","doi":"","arxiv_id":"2609.03430","authors":"Heng Wang, Jielin Qiu, Wenting Zhao, Cheng Qian, Liangwei Yang, Weizhi Zhang, Jiawei Han, Heng Ji, Silvio Savarese, Shelby Heinecke, Huan Wang","institutions":"Salesforce AI Research; University of Illinois Urbana-Champaign; University of Illinois Chicago","paper_date":"2026-09-03","venue":"arXiv (cs.CL), version 2","peer_reviewed":false,"code_url":"","data_url":"","funding_coi":"Salesforce AI Research.","figure_license":"Chart drawn by i10X from abstract and Figure 1b (CC BY 4.0)","paper_license":"CC BY 4.0","replication":"no","study_models":"Qwen3-4B\/14B\/32B; Phi-4-reasoning","study_sample":"Six reasoning tasks at ~4\u00d7 KV compression","footnotes":"","_yoast_wpseo_primary_category":""},"categories":[56,52],"tags":[],"companies":[],"products":[],"class_list":["post-6144","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-reasoning","category-studies"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Random Attention: KV Eviction Without Scoring for Reasoning - i10X News<\/title>\n<meta name=\"description\" content=\"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.\" \/>\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\/random-attention-kv-cache-eviction-reasoning\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Random Attention: KV Eviction Without Scoring for Reasoning - i10X News\" \/>\n<meta property=\"og:description\" content=\"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning\" \/>\n<meta property=\"og:site_name\" content=\"i10X News\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-19T16:49:23+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-10-19T17:21:00+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-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: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=\"2 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning\"},\"author\":{\"name\":\"Christopher Ort\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\"},\"headline\":\"Randomly Evicting KV Cache Matches Smart Selectors \u2014 and Runs 32\u201343% Faster\",\"datePublished\":\"2026-10-19T16:49:23+00:00\",\"dateModified\":\"2026-10-19T17:21:00+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning\"},\"wordCount\":417,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/random-attention-kv-cache-eviction-reasoning-title.webp\",\"articleSection\":[\"Reasoning\",\"Studies\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning\",\"name\":\"Random Attention: KV Eviction Without Scoring for Reasoning - i10X News\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/random-attention-kv-cache-eviction-reasoning-title.webp\",\"datePublished\":\"2026-10-19T16:49:23+00:00\",\"dateModified\":\"2026-10-19T17:21:00+00:00\",\"description\":\"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#primaryimage\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/random-attention-kv-cache-eviction-reasoning-title.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/random-attention-kv-cache-eviction-reasoning-title.webp\",\"width\":1344,\"height\":768,\"caption\":\"Illustration: i10X\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/random-attention-kv-cache-eviction-reasoning#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/news.i10x.ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Randomly Evicting KV Cache Matches Smart Selectors \u2014 and Runs 32\u201343% Faster\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"name\":\"i10X News\",\"description\":\"Stay updated with the latest news and announcements from i10X.\",\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/news.i10x.ai\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\",\"name\":\"i10X\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/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:\\\/\\\/news.i10x.ai\\\/#\\\/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:\\\/\\\/news.i10x.ai\\\/#\\\/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 handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been 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\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Random Attention: KV Eviction Without Scoring for Reasoning - i10X News","description":"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.","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\/random-attention-kv-cache-eviction-reasoning","og_locale":"en_US","og_type":"article","og_title":"Random Attention: KV Eviction Without Scoring for Reasoning - i10X News","og_description":"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.","og_url":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning","og_site_name":"i10X News","article_published_time":"2026-10-19T16:49:23+00:00","article_modified_time":"2026-10-19T17:21:00+00:00","og_image":[{"width":1344,"height":768,"url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-title.webp","type":"image\/webp"}],"author":"Christopher Ort","twitter_card":"summary_large_image","twitter_creator":"@i10x_ai","twitter_site":"@i10x_ai","twitter_misc":{"Written by":"Christopher Ort","Est. reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#article","isPartOf":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning"},"author":{"name":"Christopher Ort","@id":"https:\/\/news.i10x.ai\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a"},"headline":"Randomly Evicting KV Cache Matches Smart Selectors \u2014 and Runs 32\u201343% Faster","datePublished":"2026-10-19T16:49:23+00:00","dateModified":"2026-10-19T17:21:00+00:00","mainEntityOfPage":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning"},"wordCount":417,"commentCount":0,"publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"image":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-title.webp","articleSection":["Reasoning","Studies"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#respond"]}]},{"@type":"WebPage","@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning","url":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning","name":"Random Attention: KV Eviction Without Scoring for Reasoning - i10X News","isPartOf":{"@id":"https:\/\/news.i10x.ai\/#website"},"primaryImageOfPage":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#primaryimage"},"image":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-title.webp","datePublished":"2026-10-19T16:49:23+00:00","dateModified":"2026-10-19T17:21:00+00:00","description":"Random Attention matches top KV selectors on reasoning tasks and delivers 32\u201343% higher vLLM throughput by skipping scoring.","breadcrumb":{"@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#primaryimage","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-title.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/random-attention-kv-cache-eviction-reasoning-title.webp","width":1344,"height":768,"caption":"Illustration: i10X"},{"@type":"BreadcrumbList","@id":"https:\/\/i10x.ai\/news\/random-attention-kv-cache-eviction-reasoning#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/news.i10x.ai\/"},{"@type":"ListItem","position":2,"name":"Randomly Evicting KV Cache Matches Smart Selectors \u2014 and Runs 32\u201343% Faster"}]},{"@type":"WebSite","@id":"https:\/\/news.i10x.ai\/#website","url":"https:\/\/news.i10x.ai\/","name":"i10X News","description":"Stay updated with the latest news and announcements from i10X.","publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/news.i10x.ai\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/news.i10x.ai\/#organization","name":"i10X","url":"https:\/\/news.i10x.ai\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/news.i10x.ai\/#\/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:\/\/news.i10x.ai\/#\/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:\/\/news.i10x.ai\/#\/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 handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been 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"}}]}},"_links":{"self":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6144","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=6144"}],"version-history":[{"count":2,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6144\/revisions"}],"predecessor-version":[{"id":6295,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6144\/revisions\/6295"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media\/6294"}],"wp:attachment":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media?parent=6144"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/categories?post=6144"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/tags?post=6144"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/companies?post=6144"},{"taxonomy":"product","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/products?post=6144"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}