{"id":26290,"date":"2025-08-20T15:21:19","date_gmt":"2025-08-20T12:21:19","guid":{"rendered":"https:\/\/forklog.com\/en\/?p=26290"},"modified":"2025-08-22T17:43:15","modified_gmt":"2025-08-22T14:43:15","slug":"deepseek-unveils-updated-ai-model-v3-1","status":"publish","type":"post","link":"https:\/\/forklog.com\/en\/deepseek-unveils-updated-ai-model-v3-1\/","title":{"rendered":"DeepSeek Unveils Updated AI Model V3.1"},"content":{"rendered":"<p>Chinese AI startup DeepSeek has updated its flagship AI model V3 and removed the mention of the reasoning neural network R1 in its chatbot, according to <a href=\"https:\/\/www.scmp.com\/tech\/big-tech\/article\/3322481\/deepseeks-v31-update-and-missing-r1-label-spark-speculation-over-fate-r2-ai-model?module=top_story&#038;pgtype=section\">SCMP<\/a>.<\/p>\n<p>The company announced the release of V3.1 on WeChat. The update expands the model&#8217;s context window to 128,000 tokens, allowing it to retain more information during user interactions. This volume is equivalent to a book of approximately 300 pages.<\/p>\n<p>The model is also noted for its high token efficiency.<\/p>\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/lh7-qw.googleusercontent.com\/docsz\/AD_4nXdc2H3CYlEXd9c0iDC3sCDAhl4rVlprA3_t38-fCefxW5yz-OQNSxr2vXfBwRUxBAaMJZGk0ZNrKK1Kol2EGjxgM4XKpnTO2Y3V-d2GnVIofE2brAQ68q2jpUnebS4i0UTacaPB?key=g75o7QE6QQgbl7s-MBiaMw\" alt=\"DeepSeek Unveils Updated AI Model V3.1\"\/><figcaption class=\"wp-element-caption\">Source: <a href=\"https:\/\/x.com\/undefinedza\/status\/1958099010595807407\">X<\/a>.<\/figcaption><\/figure>\n<p>In the Aider Polyglot benchmark, which evaluates LLMs in solving complex programming tasks across multiple languages, DeepSeek V3.1 outperforms Claude 4 Opus.<\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">DeepSeek V3.1 beats Claude 4 Opus on Aider Polyglot<\/p>\n<p>This makes it the best non-TTC coding model and all of that for ~$1 <a href=\"https:\/\/t.co\/QyJZnVRdVK\">pic.twitter.com\/QyJZnVRdVK<\/a><\/p>\n<p>\u2014 Lisan al Gaib (@scaling01) <a href=\"https:\/\/twitter.com\/scaling01\/status\/1957890953026392212?ref_src=twsrc%5Etfw\">August 19, 2025<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>V3.1 maintains a balance between speed and quality of generation. It contains 685 billion parameters and is based on a hybrid architecture, providing high performance in tasks of dialogue, reasoning, and programming.<\/p>\n<p>DeepSeek has removed the mention of R1 from its deep thinking function. SCMP speculated that this might indicate challenges in developing the anticipated R2 version.<\/p>\n<div class=\"wp-block-text-wrappers-update article_update\"><span class=\"gtb_text-wrappers_update_head\">Update: <\/span><\/p>\n<p>On August 21, the company released an official announcement on X. <\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">Introducing DeepSeek-V3.1: our first step toward the agent era! \ud83d\ude80<\/p>\n<p>\ud83e\udde0 Hybrid inference: Think &#038; Non-Think \u2014 one model, two modes<br \/>\u26a1\ufe0f Faster thinking: DeepSeek-V3.1-Think reaches answers in less time vs. DeepSeek-R1-0528<br \/>\ud83d\udee0\ufe0f Stronger agent skills: Post-training boosts tool use and\u2026<\/p>\n<p>\u2014 DeepSeek (@deepseek_ai) <a href=\"https:\/\/twitter.com\/deepseek_ai\/status\/1958417062008918312?ref_src=twsrc%5Etfw\">August 21, 2025<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>Key features include:<\/p>\n<ul class=\"wp-block-list\">\n<li>Hybrid reasoning mode \u2014 the model decides on its own whether to engage more resources for &#8220;thinking&#8221; about a question;<\/li>\n<li>Faster thinking \u2014 V3.1 provides answers more quickly than DeepSeek-R1-0528;<\/li>\n<li>Enhanced agent skills.<\/li>\n<\/ul>\n<\/div>\n<p>The AI startup DeepSeek gained attention in January <a href=\"https:\/\/forklog.com\/en\/news\/deepseek-the-new-ai-front-runner-and-culprit-behind-cryptos-sell-off\">with the release<\/a> of the R1 model, focused on reasoning. It demonstrated high efficiency with low capital investment, leading experts to question the necessity of billion-dollar investments in the AI sector and the industry&#8217;s potential overvaluation.<\/p>\n<p>In June, the Chinese startup <a href=\"https:\/\/forklog.com\/en\/news\/deepseek-to-integrate-ai-in-chinese-hospitals\">began hiring interns<\/a> to label medical data to improve the application of artificial intelligence in hospitals.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Chinese AI startup DeepSeek has updated its flagship AI model V3 and removed the mention of the reasoning neural network R1 in its chatbot.<\/p>\n","protected":false},"author":1,"featured_media":26291,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"1","news_style_id":"1","cryptorium_level":"","_short_excerpt_text":"Chinese AI startup DeepSeek has updated its flagship AI model V3 and removed the mention of the reasoning neural network R1 in its chatbot.","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[438,1743],"class_list":["post-26290","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-artificial-intelligence","tag-deepseek"],"aioseo_notices":[],"amp_enabled":true,"views":"576","promo_type":"1","layout_type":"1","short_excerpt":"Chinese AI startup DeepSeek has updated its flagship AI model V3 and removed the mention of the reasoning neural network R1 in its chatbot.","is_update":"","_links":{"self":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/26290","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=26290"}],"version-history":[{"count":1,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/26290\/revisions"}],"predecessor-version":[{"id":26292,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/26290\/revisions\/26292"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media\/26291"}],"wp:attachment":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media?parent=26290"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/categories?post=26290"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/tags?post=26290"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}