{"id":25381,"date":"2025-01-13T17:45:27","date_gmt":"2025-01-13T15:45:27","guid":{"rendered":"https:\/\/comunicagenia.com\/blog\/sky-t1-32b-preview-arrives-the-first-open-source-reasoning-model\/"},"modified":"2025-01-13T17:45:27","modified_gmt":"2025-01-13T15:45:27","slug":"sky-t1-32b-preview-arrives-the-first-open-source-reasoning-model","status":"publish","type":"post","link":"https:\/\/comunicagenia.com\/en\/blog\/sky-t1-32b-preview-arrives-the-first-open-source-reasoning-model\/","title":{"rendered":"Sky-T1-32B-Preview arrives: the first \u201copen source\u201d reasoning model"},"content":{"rendered":"<p>Last week, NovaSky, a team of researchers at the Sky Computing Lab at the University of California, Berkeley, released Sky-T1-32B-Preview, a reasoning model that effectively competes with previous versions of the OpenAI o1 model on several key metrics.<\/p>\n<p>Sky-T1 stands out as <strong>the first truly open source reasoning model<\/strong>, as it can be replicated from scratch; The team not only published the weights of the model, but also the dataset used to train it and the code necessary for its training.<\/p>\n<p>Let&#8217;s explore the main characteristics of this model, its comparison with other similar ones and what its designation as an open source model implies.<\/p>\n<h2>Main features of the Sky-T1-32B-Preview model<\/h2>\n<p>Sky-T1-32B-Preview is a language model designed to excel in specific domains such as mathematics and programming. Some of its key features include:<\/p>\n<ul>\n<li><strong>Efficient design:<\/strong> NovaSky managed to train the model with a budget significantly lower than average for similarly sized models. This demonstrates an innovative approach to optimizing resources and training.<\/li>\n<li><strong>Competitive performance:<\/strong> Although it is in preview, Sky-T1 has proven to be able to compete with more expensive models in tests related to mathematics and coding tasks.<\/li>\n<li><strong>Access to model weights:<\/strong> NovaSky has made model weights available, making it easier for developers to experiment and build custom applications.<\/li>\n<\/ul>\n<p>Unlike some competitors, Sky-T1 offers a complete set of resources, aligning with open source principles.<\/p>\n<h2>What does it mean for a model to be open source?<\/h2>\n<p>The Open Source Initiative (OSI) recently released version 1.0 of its Open Source Artificial Intelligence (OSAID) definition. This standard establishes the requirements that a model must meet to be considered truly open:<\/p>\n<h3>Full transparency<\/h3>\n<p>An open source model should provide detailed information about its design, including training data and the code used.<\/p>\n<h3>Access and modification<\/h3>\n<p>Developers should be free to use, modify, and redistribute the model without significant restrictions.<\/p>\n<h3>Provenance of data<\/h3>\n<p>It is necessary to disclose the origin and processing of the data used in training.<\/p>\n<p>NovaSky ensures that the Sky-T1-32B-Preview meets these conditions, offering developers unprecedented access to the model&#8217;s key components.<\/p>\n<p>However, this compliance must be evaluated under the standards established by the OSI to ensure that this is not a case of &#8220;open source in name only.&#8221;<\/p>\n<h2>Comparison with other models<\/h2>\n<p>The table below shows how the Sky-T1-32B-Preview stacks up against other popular models in several key aspects. Sky-T1 stands out by offering full access to data, code and model weights, making it attractive to developers and researchers.<\/p>\n<p><a href=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&#038;w=2560\"><img title=\"Sky-T1-32B-Preview arrives: the first \u201copen source\u201d reasoning model | ComunicaGenia\" decoding=\"async\" class=\"aligncenter wp-image-16336 size-full\" src=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all\" alt=\"sky t1\" width=\"602\" height=\"278\" srcset=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all 602w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-300x139.png?strip=all 300w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-480x222.png?strip=all 480w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=60 60w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=120 120w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=180 180w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=240 240w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=361 361w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=421 421w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1.png?strip=all&amp;w=541 541w\" sizes=\"(max-width: 602px) 100vw, 602px\" \/><\/a><\/p>\n<p>In contrast, models like Journey and o1 have a more specialized approach to programming, but lack transparency in other aspects. This limits its adoption in projects that require full access to resources.<\/p>\n<p>Using a variety of tests that assess math and coding skills, we&#8217;ll examine how Sky-T1 performs against competitors such as Qwen-2.5-32B-Instruct, QwQ, and o1-preview.<\/p>\n<p><a href=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&#038;w=2560\"><img title=\"Sky-T1-32B-Preview arrives: the first \u201copen source\u201d reasoning model | ComunicaGenia\" decoding=\"async\" class=\"aligncenter wp-image-16341 size-full\" src=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all\" alt=\"sky t1 2\" width=\"759\" height=\"280\" srcset=\"https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all 759w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2-300x111.png?strip=all 300w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2-480x177.png?strip=all 480w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=75 75w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=151 151w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=227 227w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=379 379w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=531 531w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=607 607w, https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1_2.png?strip=all&amp;w=683 683w\" sizes=\"(max-width: 759px) 100vw, 759px\" \/><\/a><\/p>\n<ul>\n<li><strong>Math500:<\/strong> Sky-T1-32B-Preview scored highly (82.4), second only to QwQ (85.4). This indicates his solid ability in solving mathematical problems.<\/li>\n<li><strong>AIME2024:<\/strong> Sky-T1-32B-Preview showed a good performance (43.3), although QwQ led in this metric. Even so, Sky-T1 competes effectively with other models.<\/li>\n<li><strong>LiveCodeBench-Easy:<\/strong> Sky-T1 performed well (86.3), although it was outperformed by o1-preview (92.9) and QwQ (90.7).<\/li>\n<li><strong>LiveCodeBench-Medium:<\/strong> Sky-T1 led with a score of 56.8, outperforming QwQ and o1-preview, demonstrating its strength in medium-difficulty coding tasks.<\/li>\n<li><strong>LiveCodeBench-Hard:<\/strong> Although all scores were low, Sky-T1 led with 17.9, showing its ability in difficult coding tasks.<\/li>\n<li><strong>GPQA-Diamond<\/strong>: o1-preview dominated this test with 75.2, while Sky-T1 scored 56.8, showing good performance in generating high-quality answers.<\/li>\n<\/ul>\n<p>Sky-T1-32B-Preview demonstrates solid and competitive performance in various tests, excelling especially in medium difficulty and difficult coding tasks and advanced mathematical problems.<\/p>\n<h2>How to test Sky-T1-32B-Preview?<\/h2>\n<p>NovaSky has made several ways available to the community to experiment with Sky-T1:<\/p>\n<ul>\n<li><strong>Downloadable weights:<\/strong> Developers can access model weights from the <a href=\"https:\/\/github.com\/NovaSky-AI\/SkyThought\">NovaSky Github page<\/a>. This allows the model to be implemented in its own infrastructure.<\/li>\n<li><strong>Cloud environments:<\/strong> NovaSky has also left links to cloud platforms to test the model without the need for specialized hardware. For example, you can <a href=\"https:\/\/huggingface.co\/NovaSky-AI\">try Sky-T1 on HuggingFace<\/a>.<\/li>\n<li><strong>Complete documentation:<\/strong> The documentation includes code examples and practical applications, making the model easy to use even for beginner users.<\/li>\n<\/ul>\n<p>These options reinforce NovaSky&#8217;s commitment to democratizing AI and promoting open practices.<\/p>\n<h2>The debate on open source in AI<\/h2>\n<p>The OSI definition of open source has generated controversy in the industry. Although Sky-T1 meets key requirements, other tech giants, such as Meta, have been criticized for labeling their models as &#8220;open source&#8221; despite imposing significant restrictions.<\/p>\n<p>For example, Llama de Meta models require special licenses for large platforms, while Stability AI limits commercial use of its models.<\/p>\n<p>These cases underscore the need for clear, enforceable standards to prevent abuse of the term &#8220;open source.&#8221; The OSI hopes that the AI \u200b\u200bcommunity will adopt its definition as a reference, but there are still legal and technical challenges that could hinder its widespread application.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Last week, NovaSky, a team of researchers at the Sky Computing Lab at the University of California, Berkeley, released Sky-T1-32B-Preview, a reasoning model that effectively competes with previous versions of the OpenAI o1 model on several key metrics. Sky-T1 stands out as the first truly open source reasoning model, as it can be replicated from [&hellip;]<\/p>\n","protected":false},"author":9,"featured_media":16346,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_uag_custom_page_level_css":"","_thinkrank_schema_form_data":"","_thinkrank_selected_schema_type":"","_thinkrank_additional_schemas":"","_thinkrank_schema_disabled":"","_thinkrank_canonical_url":"","_thinkrank_og_title":"","_thinkrank_og_description":"","_thinkrank_og_image":"","_thinkrank_twitter_title":"","_thinkrank_twitter_description":"","_thinkrank_twitter_image":"","_thinkrank_imported_from":"","_thinkrank_focus_keywords":[],"_thinkrank_focus_keyword":"","_thinkrank_robots_meta_enabled":0,"_thinkrank_robots_meta":"","_thinkrank_advanced_robots_meta":"","_thinkrank_exclude_from_search":0,"_thinkrank_exclude_from_archives":0,"_thinkrank_primary_category":0,"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","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":"","ast-disable-related-posts":"","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":"","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":""}},"footnotes":""},"categories":[719],"tags":[],"class_list":["post-25381","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence"],"uagb_featured_image_src":{"full":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-scaled.jpg?strip=all",1343,768,false],"thumbnail":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-150x150.jpg?strip=all",150,150,true],"medium":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-300x171.jpg?strip=all",300,171,true],"medium_large":["https:\/\/comunicagenia.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-768x439.jpg",768,439,true],"large":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-1024x585.jpg?strip=all",1024,585,true],"1536x1536":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-1536x878.jpg?strip=all",1536,878,true],"2048x2048":["https:\/\/esaztmz2knk.exactdn.com\/wp-content\/uploads\/2025\/01\/sky-t1-32b-preview-primer-modelo-razonamiento-open-source-scaled.jpg?strip=all",1343,768,true]},"uagb_author_info":{"display_name":"marketing","author_link":"https:\/\/comunicagenia.com\/en\/blog\/author\/marketing\/"},"uagb_comment_info":0,"uagb_excerpt":"Last week, NovaSky, a team of researchers at the Sky Computing Lab at the University of California, Berkeley, released Sky-T1-32B-Preview, a reasoning model that effectively competes with previous versions of the OpenAI o1 model on several key metrics. Sky-T1 stands out as the first truly open source reasoning model, as it can be replicated from&hellip;","_links":{"self":[{"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/posts\/25381","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/comments?post=25381"}],"version-history":[{"count":0,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/posts\/25381\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/media\/16346"}],"wp:attachment":[{"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/media?parent=25381"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/categories?post=25381"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/comunicagenia.com\/en\/wp-json\/wp\/v2\/tags?post=25381"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}