{"id":22435,"date":"2025-03-26T14:06:02","date_gmt":"2025-03-26T12:06:02","guid":{"rendered":"https:\/\/forklog.com\/en\/google-unveils-reasoning-ai-model-gemini-2-5-pro\/"},"modified":"2025-03-26T14:06:02","modified_gmt":"2025-03-26T12:06:02","slug":"google-unveils-reasoning-ai-model-gemini-2-5-pro","status":"publish","type":"post","link":"https:\/\/forklog.com\/en\/google-unveils-reasoning-ai-model-gemini-2-5-pro\/","title":{"rendered":"Google Unveils &#8216;Reasoning&#8217; AI Model Gemini 2.5 Pro"},"content":{"rendered":"<p>Google has introduced a new family of &#8216;reasoning&#8217; AI models, Gemini 2.5, which pause before responding to allow for contemplation.<\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">Introducing Gemini 2.5, our most intelligent AI model. <\/p>\n<p>Our first release, an experimental version of 2.5 Pro, unlocks state-of-the-art performance in math and science. ?<\/p>\n<p>Learn more ? <a href=\"https:\/\/t.co\/aoe7egliJb\">pic.twitter.com\/aoe7egliJb<\/a><\/p>\n<p>\u2014 Google (@Google) <a href=\"https:\/\/twitter.com\/Google\/status\/1904579652816941506?ref_src=twsrc%5Etfw\">March 25, 2025<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>The first in the series is the experimental version of Gemini 2.5 Pro. This multimodal neural network with reasoning capabilities is described as &#8216;the most intelligent to date.&#8217; It is available on the Google AI Studio platform and in the Gemini app for Advanced subscription holders at $20 per month.<\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;In Gemini 2.5, we have reached a new level of performance by combining a significantly improved base model with enhanced <span data-descr=\"a stage in the machine learning process that occurs after the main model training stage is completed\" class=\"old_tooltip\">post-training<\/span>,&#8221; the corporation&#8217;s blog states.<\/p>\n<\/blockquote>\n<p>The company added that in the future, all its models will possess reasoning capabilities.<\/p>\n<p>In the Aider Polyglot test, which tracks programming skills, Gemini 2.5 Pro scored 68.6%, surpassing leading solutions from OpenAI, Anthropic, and DeepSeek.<\/p>\n<p>In the SWE-bench Verified benchmark, which measures software developers&#8217; capabilities, Google&#8217;s solution was outperformed by Claude 3.7 Sonnet\u201463.8% versus 70.3%.<\/p>\n<p>In the multimodal test Humanity\u2019s Last Exam, consisting of a thousand questions in mathematics, humanities, and natural sciences, Gemini 2.5 Pro scored 18.8%, exceeding competitors&#8217; results.<\/p>\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/lh7-qw.googleusercontent.com\/docsz\/AD_4nXeH0wEBhY788K3qbepbB1rR6UeFiqav41vzex6eSzB6m-V0odZLLZbuRkBKP7EY_ElEi_0m5mAfjyqrJaU9mTGRgSQsaA6WPq5yWDcav_42rWX535EF5S9fp_k68einMeJqvN3D?key=vLpVctNC69Bra2i8UxWa-n-J\" alt=\"Google \u0437\u0430\u043f\u0443\u0441\u0442\u0438\u043b\u0430 \u00ab\u0440\u0430\u0441\u0441\u0443\u0436\u0434\u0430\u044e\u0449\u0443\u044e\u00bb \u0418\u0418-\u043c\u043e\u0434\u0435\u043b\u044c Gemini 2.5 Pro\"\/><figcaption class=\"wp-element-caption\">Gemini 2.5 Pro&#8217;s results compared to other AI models in various benchmarks. Data: Google.<\/figcaption><\/figure>\n<p>Gemini 2.5 Pro has a context window of 1 million tokens\u2014the model can perceive approximately 750,000 words at once. There are plans to double this figure in the future.<\/p>\n<p>&#8216;Reasoning&#8217; AI requires more time and computational resources to provide accurate and comprehensive responses. The first &#8216;thinking&#8217; neural network was <a href=\"https:\/\/forklog.com\/en\/news\/openai-unveils-o1-an-ai-model-capable-of-complex-reasoning\">introduced<\/a> by OpenAI\u2014in September 2024, it released o1. Other AI developers have since embarked on creating their own solutions, including DeepSeek, <a href=\"https:\/\/forklog.com\/en\/news\/anthropics-new-hybrid-ai-model-conquers-pokemon\">Anthropic<\/a>, <a href=\"https:\/\/forklog.com\/en\/news\/alibaba-unveils-thinking-ai-model-qwq-32\">Alibaba<\/a>, and others.<\/p>\n<p>In March, Google <a href=\"https:\/\/forklog.com\/en\/news\/google-unveils-open-ai-models-for-drug-discovery\">announced<\/a> a range of AI products and features for the healthcare sector.<\/p>\n<p>Previously, the company <a href=\"https:\/\/forklog.com\/en\/news\/googles-gemini-chatbot-learns-to-analyze-search-history\">trained<\/a> its AI assistant Gemini to analyze and consider a user&#8217;s search history to provide more relevant responses.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Google has introduced a new family of &#8216;reasoning&#8217; AI models, Gemini 2.5, which pause before responding to allow for contemplation. Introducing Gemini 2.5, our most intelligent AI model. Our first release, an experimental version of 2.5 Pro, unlocks state-of-the-art performance in math and science. ? Learn more ? pic.twitter.com\/aoe7egliJb \u2014 Google (@Google) March 25, 2025 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":22434,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"","news_style_id":"","cryptorium_level":"","_short_excerpt_text":"","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[438,719,738],"class_list":["post-22435","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-artificial-intelligence","tag-gemini","tag-google"],"aioseo_notices":[],"amp_enabled":true,"views":"104","promo_type":"","layout_type":"","short_excerpt":"","is_update":"","_links":{"self":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/22435","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/comments?post=22435"}],"version-history":[{"count":0,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/22435\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media\/22434"}],"wp:attachment":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media?parent=22435"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/categories?post=22435"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/tags?post=22435"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}