• Home
  • Services
  • About
  • Contact
  • Blog
  • 知財活動のROICへの貢献
  • 生成AIを活用した知財戦略の策定方法
  • 生成AIとの「壁打ち」で、新たな発明を創出する方法

​
​よろず知財コンサルティングのブログ

国産フルスクラッチLLM「PLaMo 3.0 Prime」

28/3/2026

0 Comments

 
2026年3月19日、株式会社Preferred Networksは日本初の「長考(推論)」機能を備えた国産生成AI基盤モデル「PLaMo 3.0 Prime」を発表しました。
既存モデルを基盤とせずゼロから開発された「フルスクラッチ」モデルであり、国内で開発されたものとしては初めて、複雑な指示に対して段階的に思考を重ねて質の高い回答を導き出す「Reasoning(長考)」機能を搭載した点で画期的と言え、日本語の対話や指示追従能力において国際的な競合モデルに匹敵する性能を示しています。
現在は2026年6月の商用版提供に向け、実運用での負荷や応答性能を検証するためのモニター企業募集が開始されています。
生成AIに「PLaMo 3.0 Prime」について深堀させました。なお、生成AIによる調査・分析結果は、公開された情報からだけの分析であり、必ずしも実情を示したものではないこと、誤った情報も含まれていることについてはご留意されたうえで、ご参照ください。
 
初の“長考”できる国産フルスクラッチLLM「PLaMo 3.0 Prime」 Qwen3-235Bやgpt-oss-120bに肉薄 PFN
2026年03月23日
https://www.itmedia.co.jp/aiplus/articles/2603/23/news112.html
 
生成AI基盤モデルPLaMo 3.0 Primeβ版のモニター企業募集
フルスクラッチ開発の国産生成AIで初となる、論理的思考に優れ、複雑なタスクに対応するReasoningモデル
https://www.preferred.jp/ja/news/pr20260319
 
 
Japan’s Fully Scratch-Built LLM “PLaMo 3.0 Prime”
On March 19, 2026, Preferred Networks announced “PLaMo 3.0 Prime,” a domestically developed generative AI foundation model equipped with Japan’s first “long-thinking (reasoning)” capability.
Unlike models built on top of existing architectures, PLaMo 3.0 Prime is a fully scratch-built model developed from the ground up. It is considered groundbreaking as the first domestically developed model to incorporate a “Reasoning (long-thinking)” function that enables step-by-step thinking over complex instructions to produce high-quality responses. It also demonstrates performance in Japanese dialogue and instruction-following that is competitive with leading international models.
Currently, in preparation for its commercial release scheduled for June 2026, the company has begun recruiting monitor companies to evaluate system load and response performance in real-world operational environments.
I used generative AI to conduct a deeper analysis of “PLaMo 3.0 Prime.” Please note that the results of this analysis are based solely on publicly available information and may not necessarily reflect actual conditions. They may also contain inaccuracies.

Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
Your browser does not support viewing this document. Click here to download the document.
0 Comments



Leave a Reply.

    著者

    萬秀憲

    アーカイブ

    January 2026
    December 2025
    November 2025
    October 2025
    September 2025
    August 2025
    July 2025
    June 2025
    May 2025
    April 2025
    March 2025
    February 2025
    January 2025
    December 2024
    November 2024
    October 2024
    September 2024
    August 2024
    July 2024
    June 2024
    May 2024
    April 2024
    March 2024
    February 2024
    January 2024
    December 2023
    November 2023
    October 2023
    September 2023
    August 2023
    July 2023
    June 2023
    May 2023
    April 2023
    March 2023
    February 2023
    January 2023
    December 2022
    November 2022
    October 2022
    September 2022
    August 2022
    July 2022
    June 2022
    May 2022
    April 2022
    March 2022
    February 2022
    January 2022
    December 2021
    November 2021
    October 2021
    September 2021
    August 2021
    July 2021
    June 2021
    May 2021
    April 2021
    March 2021
    February 2021
    January 2021
    December 2020
    November 2020
    October 2020
    September 2020
    August 2020
    July 2020
    June 2020

    カテゴリー

    All

    RSS Feed

Copyright © よろず知財戦略コンサルティング All Rights Reserved.
サイトはWeeblyにより提供され、お名前.comにより管理されています
  • Home
  • Services
  • About
  • Contact
  • Blog
  • 知財活動のROICへの貢献
  • 生成AIを活用した知財戦略の策定方法
  • 生成AIとの「壁打ち」で、新たな発明を創出する方法