Nippon Telegraph and Telephone (NTT) Corp., a Japanese telecommunications company, has announced the development of a large language model (LLM) named tsuzumi. The LLM, scheduled for release in March, is intended to become the basis for AI learning Japanese.

tsuzumi is an independently developed AI model based on significant research in natural language processing. Its key features include industry-specific customization, cost-effectiveness, and availability in two versions with different learning scale parameters–one with 600 million and the other with 7 billion.

According to NTT, the LLM can reduce operational costs and power consumption, owing to the industry-focused tailorability of tsuzumi. In addition, the model helps in addressing specific industry needs more effectively.

Analyst QuickTake: This news comes just a couple of days after Japanese conglomerate SoftBank unveiled a GenAI computing platform intended to help in the development of LLMs geared toward the Japanese language, and AIST and Tokyo Tech initiated a project to develop a 1 75 billion-parameter Japanese LLM.