{"id":33393,"date":"2025-04-17T12:36:30","date_gmt":"2025-04-17T04:36:30","guid":{"rendered":"https:\/\/www.1ai.net\/?p=33393"},"modified":"2025-04-17T12:36:30","modified_gmt":"2025-04-17T04:36:30","slug":"%e5%ad%97%e8%8a%82%e8%b7%b3%e5%8a%a8%e5%8f%91%e5%b8%83%e8%b1%86%e5%8c%85-1-5-%e6%b7%b1%e5%ba%a6%e6%80%9d%e8%80%83%e6%a8%a1%e5%9e%8b%ef%bc%8c%e5%85%b7%e5%a4%87%e7%9c%8b%e5%9b%be%e6%80%9d","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/33393.html","title":{"rendered":"ByteDance Releases Beanbag 1.5 Deep Thinking Model with \"Thinking in Pictures\" Capability"},"content":{"rendered":"

April 17, 2011 - At the site of today's Volcano Engine AI Innovation Tour in Hangzhou, theByteDance<\/a>The president of Volcano Engine, Tan Toi, has released his latestBean curd<\/a> 1.5\u30fbdeep thinking model<\/a>.<\/p>\n

\"d6be4344j00suui41001hd000j600c9p\"<\/p>\n

According to the report, the model excelled in specialized areas such as math, programming, scientific reasoning and other general-purpose tasks such as creative writing. The mathematical reasoning AIME 2024 test score tied OpenAI o3-mini-high, and the programming competition and scientific reasoning test scores were close to o1. In general tasks such as creative writing and humanities quiz, the model also demonstrated excellent generalization ability, and is capable of a wider range of use scenarios.<\/p>\n

The technical report shows that the model adopts MoE architecture with 200B total parameters and 20B activation parameters, which is lower than the industry's similar model parameter scale of 50%, and has a significant reasoning cost advantage. The beanbag model API service guarantees high concurrency with latency as low as 20 milliseconds.<\/p>\n

At the conference, Tan Yu introduced the powerful features of the Beanbag 1.5 Deep Thinking Model. The model can be used in conjunction with visual understanding technology to provide more functionality, such as analyzing landforms based on photographs, or assisting in ordering food while traveling, or completing enterprise project management flowcharts.<\/p>\n

In addition, Doubao literacy model Seedream3.0 is also officially open to enterprises. The model has been on line in the Dream AI, Doubao products, and recently appeared on the first echelon of the list of authoritative arena Artificial Analysis text-generated charts.Seedream3.0 supports 2K resolution straight out, and has advantages in terms of the accuracy of the structure of the birth charts, the generation of small words and layout, aesthetics, and realism.<\/p>\n

Of interest is that the new model also has better video search capabilities, with the model being able to find the appropriate answer in the video based on the user's search question.<\/p>\n

Meanwhile, 1AI noted that Tan Stay said that the beanbag big model usage continues to grow at a high rate. As of March 2025, the average daily tokens usage of the beanbag big model exceeded 12.7 trillion, an increase of more than 106 times compared to the beginning of the release.<\/p>","protected":false},"excerpt":{"rendered":"

April 17 news, in today's volcano engine AI innovation tour exhibition Hangzhou station site, byte jumping's volcano engine president Tan to be released the latest beanbag 1.5\u30fbdeep thinking model. According to reports, the model has outstanding performance in professional fields such as math, programming, scientific reasoning, and general tasks such as creative writing. In mathematical reasoning AIME 2024 test, the model scored equal to OpenAI o3-mini-high, and in programming competition and scientific reasoning test, the model scored close to o1. In general tasks such as creative writing and humanities quiz, the model also demonstrated excellent generalization ability, which is able to cope with a wider range of usage scenarios. According to the technical report, the model adopts MoE architecture, with 200B total parameters and 20B activation parameters, which is lower than that of similar models in the industry.<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[548,5744,2248],"collection":[],"class_list":["post-33393","post","type-post","status-publish","format-standard","hentry","category-news","tag-548","tag-5744","tag-2248"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/33393","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=33393"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/33393\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=33393"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=33393"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=33393"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=33393"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}