Japan’s government-backed push to develop sovereign AI capability produced a concrete output this week.
Rakuten Group released Rakuten AI 3.0 on March 17, completing the company’s development work under Japan’s GENIAC program. The Generative AI Accelerator Challenge is promoted jointly by Japan’s Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO), a structure that puts Rakuten AI 3.0 in a distinct category from commercially-initiated model releases.
According to Rakuten’s official press room, the model was developed as part of the program’s third term, with Rakuten selected specifically to build Japanese language-optimized AI. Rakuten describes AI 3.0 as Japan’s largest high-performance AI model, that’s the company’s own designation, and no independent benchmark comparison against other Japanese LLMs has confirmed it.
The target audience is companies and developers building AI applications for Japanese-language contexts. Rakuten’s GENIAC selection announcement makes clear the model was built with Japanese-language optimization as its core design constraint, not a post-hoc localization of a general-purpose English model.
Model architecture, parameter count, API access terms, and pricing haven’t been confirmed in available materials. The Wire’s source schema for this item was delivered incomplete. A follow-up brief will cover technical details once the complete schema is available.