Rakuten Group Inc. has unveiled “Rakuten AI 3.0,” a groundbreaking Japanese large language model (LLM) with about 700 billion parameters, making it one of the most significant models in Japan. It is designed to enhance generative artificial intelligence with a keen understanding of Japanese language nuances, culture, and customs. This innovative model aims to be integrated gradually into Rakuten’s online shopping services and various platforms, dramatically reducing operational costs by up to 90% compared to other LLMs of similar scale. Additionally, the model is expected to be available for external use by spring 2026. The development of Rakuten AI 3.0 was facilitated by the “Geniac” initiative from the Economy, Trade and Industry Ministry, which supports domestic AI advancements through subsidies for necessary technologies. Competitors like NTT Inc. and a SoftBank subsidiary are also in the LLM race, with SoftBank’s “Sarashina” model achieving approximately 460 billion parameters.
Source link
Share
Read more