Rakuten said the project aims to support the development of generative AI systems in the Asian nation through the provision of computing resources and technical exchange
In sum – what to know:
Rakuten joins GENIAC Phase 3 – The company was selected for a government-backed program supporting generative AI R&D with compute access and collaboration.
Focus on open-weight Japanese LLM – Rakuten aims to improve model memory and personalization by developing a lightweight, memory-augmented Japanese language model using a Mixture of Experts architecture.
Plans for broader applications – The LLM will serve as a foundation for personalized AI tools across Rakuten’s services, with a focus on long-term interaction, efficiency and language-specific performance.
Japanese giant Rakuten Group has been selected for the third phase of the Generative AI Accelerator Challenge (GENIAC), a program supported by Japan’s Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO), the former said in a release.
The project aims to support the development of generative AI systems in the Asian nation through the provision of computing resources and technical exchange.
The GENIAC initiative began in February 2024 and is structured in phases. Rakuten was selected through the application process for the program’s third phase, which started in March 2025. The third phase of GENIAC focuses on improving domestic research capacity for large language models (LLMs), including infrastructure and algorithmic advancements, the company said.
The company also noted it plans to start research and development in August 2025. Its proposed work will focus on creating a Japanese language foundation model with open weights. The model will incorporate techniques to expand memory capacity, improving its ability to retain and retrieve information across longer prompts and interactions. According to the company, this is intended to address existing technical constraints in generative models that operate with limited contextual memory.
Rakuten has previously released Japanese-language models to the open-source community, including Rakuten AI 2.0, which employs a Mixture of Experts (MoE) architecture. This model architecture activates only relevant components during inference, reducing computational overhead compared to standard dense models, the company said.
In the new phase of development, the Japanese group intends to test enhancements that would enable models to handle extended interactions and deliver more consistent outputs across multiple user prompts.
In the longer term, the company aims to develop LLMs capable of retaining user-specific context to enable more tailored interactions. This includes investigating training and inference techniques to support personalized AI applications. These models are intended for future use across Rakuten’s various platforms, with a focus on automation and customer interaction support.
Yu Hirate, vice general manager at Rakuten Group’s AI research supervisory department, stated: “I am very pleased to be able to work on the development of a cutting-edge generative AI foundation model with the support of NEDO and the Ministry of Economy, Trade and Industry. Through this cost-effective AI model, we hope to contribute to the realization of AI agents that are best optimized for the Japanese language and are highly personalized, as well as empower local businesses and boost the economy.”
In January, Rakuten Mobile, part of the Rakuten Group, announced the launch of “Rakuten AI for Business,” a Generative AI solution designed to assist corporate clients in various business functions.
With advanced language understanding and task automation capabilities, the service provides quick and relevant responses to user queries. It supports activities such as document generation, translation, idea development, analysis and research through an interactive chat interface, Rakuten Mobile said.
Rakuten Mobile also noted its new AI tool is specifically tailored to align with Japanese business culture, regulations and industry norms.
The carrier also noted that Rakuten AI for Business operates in a secure environment, ensuring that user inputs are not utilized for AI model training without explicit permission. It also allows companies to manage sensitive data by restricting specific keywords from being processed by the AI.
