{"id":41721,"date":"2025-10-02T06:48:47","date_gmt":"2025-10-02T06:48:47","guid":{"rendered":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/"},"modified":"2025-10-02T06:48:47","modified_gmt":"2025-10-02T06:48:47","slug":"servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget","status":"publish","type":"post","link":"https:\/\/youzum.net\/it\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/","title":{"rendered":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget"},"content":{"rendered":"<p>ServiceNow AI Research Lab has released <strong>Apriel-1.5-15B-Thinker<\/strong>, a 15-billion-parameter open-weights multimodal reasoning model trained with a data-centric <strong>mid-training<\/strong> recipe\u2014continual pretraining followed by supervised fine-tuning\u2014<strong>without<\/strong> reinforcement learning or preference optimization. The model attains an Artificial Analysis Intelligence Index score of 52 with 8x cost savings compared to SOTA. The checkpoint ships under an <a href=\"https:\/\/huggingface.co\/ServiceNow-AI\/Apriel-1.5-15b-Thinker\" target=\"_blank\" rel=\"noreferrer noopener\">MIT license on Hugging Face.<\/a><\/p>\n<h3 class=\"wp-block-heading\"><strong>So, What\u2019s new in it for me?<\/strong><\/h3>\n<ul class=\"wp-block-list\">\n<li><strong>Frontier-level composite score at small scale.<\/strong> The model reports <strong>Artificial Analysis Intelligence Index (AAI)<\/strong> = <strong>52<\/strong>, matching <strong>DeepSeek-R1-0528<\/strong> on that combined metric while being dramatically smaller. AAI aggregates 10 third-party evaluations (MMLU-Pro, GPQA Diamond, Humanity\u2019s Last Exam, LiveCodeBench, SciCode, AIME 2025, IFBench, AA-LCR, Terminal-Bench Hard, \u03c4\u00b2-Bench Telecom).<\/li>\n<li><strong>Single-GPU deployability.<\/strong> The model card states the 15B checkpoint \u201cfits on a single GPU,\u201d targeting on-premises and air-gapped deployments with fixed memory and latency budgets.<\/li>\n<li><strong>Open weights and reproducible pipeline.<\/strong> Weights, training recipe, and evaluation protocol are public for independent verification. <\/li>\n<\/ul>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large is-resized\"><img fetchpriority=\"high\" decoding=\"async\" width=\"1024\" height=\"582\" data-attachment-id=\"75017\" data-permalink=\"https:\/\/www.marktechpost.com\/2025\/10\/01\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/screenshot-2025-10-01-at-9-23-02-pm-2\/\" data-orig-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1.png\" data-orig-size=\"2018,1146\" data-comments-opened=\"1\" data-image-meta='{\"aperture\":\"0\",\"credit\":\"\",\"camera\":\"\",\"caption\":\"\",\"created_timestamp\":\"0\",\"copyright\":\"\",\"focal_length\":\"0\",\"iso\":\"0\",\"shutter_speed\":\"0\",\"title\":\"\",\"orientation\":\"0\"}' data-image-title=\"Screenshot 2025-10-01 at 9.23.02\u202fPM\" data-image-description=\"\" data-image-caption=\"\" data-medium-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-300x170.png\" data-large-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582.png\" src=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582.png\" alt=\"\" class=\"wp-image-75017\" \/><figcaption class=\"wp-element-caption\">https:\/\/huggingface.co\/ServiceNow-AI\/Apriel-1.5-15b-Thinker<\/figcaption><\/figure>\n<\/div>\n<h3 class=\"wp-block-heading\"><strong>Ok! I got it but what is it\u2019s training mechanism?<\/strong><\/h3>\n<p><strong>Base and upscaling.<\/strong> Apriel-1.5-15B-Thinker starts from <strong>Mistral\u2019s Pixtral-12B-Base-2409<\/strong> multimodal decoder-vision stack. The research team applies <strong>depth upscaling<\/strong>\u2014increasing decoder layers from 40\u219248\u2014then <strong>projection-network realignment<\/strong> to align the vision encoder with the enlarged decoder. This avoids pretraining from scratch while preserving single-GPU deployability.<\/p>\n<p><strong>CPT (Continual Pretraining).<\/strong> Two stages: (1) mixed text+image data to build foundational reasoning and document\/diagram understanding; (2) targeted synthetic visual tasks (reconstruction, matching, detection, counting) to sharpen spatial and compositional reasoning. Sequence lengths extend to 32k and 16k tokens respectively, with selective loss placement on response tokens for instruction-formatted samples.<\/p>\n<p><strong>SFT (Supervised Fine-Tuning).<\/strong> High-quality, reasoning-trace instruction data for math, coding, science, tool use, and instruction following; two additional SFT runs (stratified subset; longer-context) are <strong>weight-merged<\/strong> to form the final checkpoint. No RL (reinforcement learning) or RLAIF (reinforcement learning from AI feedback).<\/p>\n<p><strong>Data note.<\/strong> <strong>~25% of the depth-upscaling text mix<\/strong> derives from <a href=\"https:\/\/www.linkedin.com\/posts\/nvidia-ai_congratulations-to-servicenow-ai-research-activity-7378866427732291584-N9DQ\/?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>NVIDIA\u2019s Nemotron<\/strong> collection<\/a>.<\/p>\n<h3 class=\"wp-block-heading\"><strong>O\u2019 Wow! Tell me about it\u2019s results then?<\/strong><\/h3>\n<p><strong>Key text benchmarks (pass@1 \/ accuracy).<\/strong><\/p>\n<ul class=\"wp-block-list\">\n<li><strong>AIME 2025 (American Invitational Mathematics Examination 2025):<\/strong> <strong>87.5\u201388%<\/strong><\/li>\n<li><strong>GPQA Diamond (Graduate-Level Google-Proof Question Answering, Diamond split):<\/strong> <strong>\u224871%<\/strong> <\/li>\n<li><strong>IFBench (Instruction-Following Benchmark):<\/strong> <strong>~62<\/strong><\/li>\n<li><strong>\u03c4\u00b2-Bench (Tau-squared Bench) Telecom:<\/strong> <strong>~68<\/strong><\/li>\n<li><strong>LiveCodeBench (functional code correctness):<\/strong> <strong>~72.8<\/strong><\/li>\n<\/ul>\n<p>Using <strong>VLMEvalKit<\/strong> for reproducibility, Apriel scores competitively across <strong>MMMU \/ MMMU-Pro (Massive Multi-discipline Multimodal Understanding), LogicVista, MathVision, MathVista, MathVerse, MMStar, CharXiv, AI2D, BLINK<\/strong>, with stronger results on documents\/diagrams and text-dominant math imagery.<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large is-resized\"><img decoding=\"async\" width=\"1024\" height=\"601\" data-attachment-id=\"75019\" data-permalink=\"https:\/\/www.marktechpost.com\/2025\/10\/01\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/screenshot-2025-10-01-at-9-28-17-pm-2\/\" data-orig-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.28.17-PM-1.png\" data-orig-size=\"1902,1116\" data-comments-opened=\"1\" data-image-meta='{\"aperture\":\"0\",\"credit\":\"\",\"camera\":\"\",\"caption\":\"\",\"created_timestamp\":\"0\",\"copyright\":\"\",\"focal_length\":\"0\",\"iso\":\"0\",\"shutter_speed\":\"0\",\"title\":\"\",\"orientation\":\"0\"}' data-image-title=\"Screenshot 2025-10-01 at 9.28.17\u202fPM\" data-image-description=\"\" data-image-caption=\"\" data-medium-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.28.17-PM-1-300x176.png\" data-large-file=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.28.17-PM-1-1024x601.png\" src=\"https:\/\/www.marktechpost.com\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.28.17-PM-1-1024x601.png\" alt=\"\" class=\"wp-image-75019\" \/><figcaption class=\"wp-element-caption\">https:\/\/huggingface.co\/ServiceNow-AI\/Apriel-1.5-15b-Thinker\/blob\/main\/Apriel-1.5-Thinker.pdf<\/figcaption><\/figure>\n<\/div>\n<h3 class=\"wp-block-heading\"><strong>Lets Summarize everything<\/strong><\/h3>\n<p>Apriel-1.5-15B-Thinker demonstrates that careful mid-training (continual pretraining + supervised fine-tuning, no reinforcement learning) can deliver a 52 on the Artificial Analysis Intelligence Index (AAI) while remaining deployable on a single graphics processing unit. Reported task-level scores (for example, AIME 2025 \u224888, GPQA Diamond \u224871, IFBench \u224862, Tau-squared Bench Telecom \u224868) align with the model card and place the 15-billion-parameter checkpoint in the most cost-efficient band of current open-weights reasoners. For enterprises, that combination\u2014open weights, reproducible recipe, and single-GPU latency\u2014makes Apriel a practical baseline to evaluate before considering larger closed systems.<\/p>\n<p>The post <a href=\"https:\/\/www.marktechpost.com\/2025\/10\/01\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\">ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget<\/a> appeared first on <a href=\"https:\/\/www.marktechpost.com\/\">MarkTechPost<\/a>.<\/p>","protected":false},"excerpt":{"rendered":"<p>ServiceNow AI Research Lab has released Apriel-1.5-15B-Thinker, a 15-billion-parameter open-weights multimodal reasoning model trained with a data-centric mid-training recipe\u2014continual pretraining followed by supervised fine-tuning\u2014without reinforcement learning or preference optimization. The model attains an Artificial Analysis Intelligence Index score of 52 with 8x cost savings compared to SOTA. The checkpoint ships under an MIT license on Hugging Face. So, What\u2019s new in it for me? Frontier-level composite score at small scale. The model reports Artificial Analysis Intelligence Index (AAI) = 52, matching DeepSeek-R1-0528 on that combined metric while being dramatically smaller. AAI aggregates 10 third-party evaluations (MMLU-Pro, GPQA Diamond, Humanity\u2019s Last Exam, LiveCodeBench, SciCode, AIME 2025, IFBench, AA-LCR, Terminal-Bench Hard, \u03c4\u00b2-Bench Telecom). Single-GPU deployability. The model card states the 15B checkpoint \u201cfits on a single GPU,\u201d targeting on-premises and air-gapped deployments with fixed memory and latency budgets. Open weights and reproducible pipeline. Weights, training recipe, and evaluation protocol are public for independent verification. https:\/\/huggingface.co\/ServiceNow-AI\/Apriel-1.5-15b-Thinker Ok! I got it but what is it\u2019s training mechanism? Base and upscaling. Apriel-1.5-15B-Thinker starts from Mistral\u2019s Pixtral-12B-Base-2409 multimodal decoder-vision stack. The research team applies depth upscaling\u2014increasing decoder layers from 40\u219248\u2014then projection-network realignment to align the vision encoder with the enlarged decoder. This avoids pretraining from scratch while preserving single-GPU deployability. CPT (Continual Pretraining). Two stages: (1) mixed text+image data to build foundational reasoning and document\/diagram understanding; (2) targeted synthetic visual tasks (reconstruction, matching, detection, counting) to sharpen spatial and compositional reasoning. Sequence lengths extend to 32k and 16k tokens respectively, with selective loss placement on response tokens for instruction-formatted samples. SFT (Supervised Fine-Tuning). High-quality, reasoning-trace instruction data for math, coding, science, tool use, and instruction following; two additional SFT runs (stratified subset; longer-context) are weight-merged to form the final checkpoint. No RL (reinforcement learning) or RLAIF (reinforcement learning from AI feedback). Data note. ~25% of the depth-upscaling text mix derives from NVIDIA\u2019s Nemotron collection. O\u2019 Wow! Tell me about it\u2019s results then? Key text benchmarks (pass@1 \/ accuracy). AIME 2025 (American Invitational Mathematics Examination 2025): 87.5\u201388% GPQA Diamond (Graduate-Level Google-Proof Question Answering, Diamond split): \u224871% IFBench (Instruction-Following Benchmark): ~62 \u03c4\u00b2-Bench (Tau-squared Bench) Telecom: ~68 LiveCodeBench (functional code correctness): ~72.8 Using VLMEvalKit for reproducibility, Apriel scores competitively across MMMU \/ MMMU-Pro (Massive Multi-discipline Multimodal Understanding), LogicVista, MathVision, MathVista, MathVerse, MMStar, CharXiv, AI2D, BLINK, with stronger results on documents\/diagrams and text-dominant math imagery. https:\/\/huggingface.co\/ServiceNow-AI\/Apriel-1.5-15b-Thinker\/blob\/main\/Apriel-1.5-Thinker.pdf Lets Summarize everything Apriel-1.5-15B-Thinker demonstrates that careful mid-training (continual pretraining + supervised fine-tuning, no reinforcement learning) can deliver a 52 on the Artificial Analysis Intelligence Index (AAI) while remaining deployable on a single graphics processing unit. Reported task-level scores (for example, AIME 2025 \u224888, GPQA Diamond \u224871, IFBench \u224862, Tau-squared Bench Telecom \u224868) align with the model card and place the 15-billion-parameter checkpoint in the most cost-efficient band of current open-weights reasoners. For enterprises, that combination\u2014open weights, reproducible recipe, and single-GPU latency\u2014makes Apriel a practical baseline to evaluate before considering larger closed systems. The post ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget appeared first on MarkTechPost.<\/p>","protected":false},"author":2,"featured_media":41722,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"pmpro_default_level":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"_pvb_checkbox_block_on_post":false,"footnotes":""},"categories":[52,5,7,1],"tags":[],"class_list":["post-41721","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-club","category-committee","category-news","category-uncategorized","pmpro-has-access"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v25.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum<\/title>\n<meta name=\"description\" content=\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/youzum.net\/it\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\" \/>\n<meta property=\"og:locale\" content=\"it_IT\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum\" \/>\n<meta property=\"og:description\" content=\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\" \/>\n<meta property=\"og:url\" content=\"https:\/\/youzum.net\/it\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\" \/>\n<meta property=\"og:site_name\" content=\"YouZum\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/DroneAssociationTH\/\" \/>\n<meta property=\"article:published_time\" content=\"2025-10-02T06:48:47+00:00\" \/>\n<meta name=\"author\" content=\"admin NU\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Scritto da\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin NU\" \/>\n\t<meta name=\"twitter:label2\" content=\"Tempo di lettura stimato\" \/>\n\t<meta name=\"twitter:data2\" content=\"3 minuti\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\"},\"author\":{\"name\":\"admin NU\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c\"},\"headline\":\"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget\",\"datePublished\":\"2025-10-02T06:48:47+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\"},\"wordCount\":539,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\"},\"image\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp\",\"articleSection\":[\"AI\",\"Committee\",\"News\",\"Uncategorized\"],\"inLanguage\":\"it-IT\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\",\"url\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\",\"name\":\"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum\",\"isPartOf\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp\",\"datePublished\":\"2025-10-02T06:48:47+00:00\",\"description\":\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\",\"breadcrumb\":{\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#breadcrumb\"},\"inLanguage\":\"it-IT\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"it-IT\",\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage\",\"url\":\"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp\",\"contentUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp\",\"width\":1024,\"height\":582},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/youzum.net\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/yousum.gpucore.co\/#website\",\"url\":\"https:\/\/yousum.gpucore.co\/\",\"name\":\"YouSum\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/yousum.gpucore.co\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"it-IT\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\",\"name\":\"Drone Association Thailand\",\"url\":\"https:\/\/yousum.gpucore.co\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"it-IT\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png\",\"contentUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png\",\"width\":300,\"height\":300,\"caption\":\"Drone Association Thailand\"},\"image\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/DroneAssociationTH\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c\",\"name\":\"admin NU\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"it-IT\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png\",\"contentUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png\",\"caption\":\"admin NU\"},\"url\":\"https:\/\/youzum.net\/it\/members\/adminnu\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum","description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","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:\/\/youzum.net\/it\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/","og_locale":"it_IT","og_type":"article","og_title":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum","og_description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","og_url":"https:\/\/youzum.net\/it\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/","og_site_name":"YouZum","article_publisher":"https:\/\/www.facebook.com\/DroneAssociationTH\/","article_published_time":"2025-10-02T06:48:47+00:00","author":"admin NU","twitter_card":"summary_large_image","twitter_misc":{"Scritto da":"admin NU","Tempo di lettura stimato":"3 minuti"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#article","isPartOf":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/"},"author":{"name":"admin NU","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c"},"headline":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget","datePublished":"2025-10-02T06:48:47+00:00","mainEntityOfPage":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/"},"wordCount":539,"commentCount":0,"publisher":{"@id":"https:\/\/yousum.gpucore.co\/#organization"},"image":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage"},"thumbnailUrl":"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp","articleSection":["AI","Committee","News","Uncategorized"],"inLanguage":"it-IT","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/","url":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/","name":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget - YouZum","isPartOf":{"@id":"https:\/\/yousum.gpucore.co\/#website"},"primaryImageOfPage":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage"},"image":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage"},"thumbnailUrl":"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp","datePublished":"2025-10-02T06:48:47+00:00","description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","breadcrumb":{"@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#breadcrumb"},"inLanguage":"it-IT","potentialAction":[{"@type":"ReadAction","target":["https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/"]}]},{"@type":"ImageObject","inLanguage":"it-IT","@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#primaryimage","url":"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp","contentUrl":"https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp","width":1024,"height":582},{"@type":"BreadcrumbList","@id":"https:\/\/youzum.net\/servicenow-ai-releases-apriel-1-5-15b-thinker-an-open-weights-multimodal-reasoning-model-that-hits-frontier-level-performance-on-a-single-gpu-budget\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/youzum.net\/"},{"@type":"ListItem","position":2,"name":"ServiceNow AI Releases Apriel-1.5-15B-Thinker: An Open-Weights Multimodal Reasoning Model that Hits Frontier-Level Performance on a Single-GPU Budget"}]},{"@type":"WebSite","@id":"https:\/\/yousum.gpucore.co\/#website","url":"https:\/\/yousum.gpucore.co\/","name":"YouSum","description":"","publisher":{"@id":"https:\/\/yousum.gpucore.co\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/yousum.gpucore.co\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"it-IT"},{"@type":"Organization","@id":"https:\/\/yousum.gpucore.co\/#organization","name":"Drone Association Thailand","url":"https:\/\/yousum.gpucore.co\/","logo":{"@type":"ImageObject","inLanguage":"it-IT","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/","url":"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png","contentUrl":"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png","width":300,"height":300,"caption":"Drone Association Thailand"},"image":{"@id":"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/DroneAssociationTH\/"]},{"@type":"Person","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c","name":"admin NU","image":{"@type":"ImageObject","inLanguage":"it-IT","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/image\/","url":"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png","contentUrl":"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png","caption":"admin NU"},"url":"https:\/\/youzum.net\/it\/members\/adminnu\/"}]}},"rttpg_featured_image_url":{"full":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"landscape":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"portraits":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"thumbnail":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-150x150.webp",150,150,true],"medium":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-300x171.webp",300,171,true],"large":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"1536x1536":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"2048x2048":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD.webp",1024,582,false],"trp-custom-language-flag":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-18x10.webp",18,10,true],"woocommerce_thumbnail":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-300x300.webp",300,300,true],"woocommerce_single":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-600x341.webp",600,341,true],"woocommerce_gallery_thumbnail":["https:\/\/youzum.net\/wp-content\/uploads\/2025\/10\/Screenshot-2025-10-01-at-9.23.02-PM-1-1024x582-VDKoOD-100x100.webp",100,100,true]},"rttpg_author":{"display_name":"admin NU","author_link":"https:\/\/youzum.net\/it\/members\/adminnu\/"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/youzum.net\/it\/category\/ai-club\/\" rel=\"category tag\">AI<\/a> <a href=\"https:\/\/youzum.net\/it\/category\/committee\/\" rel=\"category tag\">Committee<\/a> <a href=\"https:\/\/youzum.net\/it\/category\/news\/\" rel=\"category tag\">News<\/a> <a href=\"https:\/\/youzum.net\/it\/category\/uncategorized\/\" rel=\"category tag\">Uncategorized<\/a>","rttpg_excerpt":"ServiceNow AI Research Lab has released Apriel-1.5-15B-Thinker, a 15-billion-parameter open-weights multimodal reasoning model trained with a data-centric mid-training recipe\u2014continual pretraining followed by supervised fine-tuning\u2014without reinforcement learning or preference optimization. The model attains an Artificial Analysis Intelligence Index score of 52 with 8x cost savings compared to SOTA. The checkpoint ships under an MIT license on&hellip;","_links":{"self":[{"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/posts\/41721","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/comments?post=41721"}],"version-history":[{"count":0,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/posts\/41721\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/media\/41722"}],"wp:attachment":[{"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/media?parent=41721"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/categories?post=41721"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/youzum.net\/it\/wp-json\/wp\/v2\/tags?post=41721"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}