PLaMo Embedding 1B (JP)

PLaMo Embedding 1B (JP) is a Japanese text embedding model from Preferred Networks, specifically optimized for generating high-quality vector representations of Japanese text. Built with 1 billion parameters, it delivers strong embedding quality for Japanese semantic search, retrieval, and similarity tasks.

The model handles the unique characteristics of Japanese text including multiple writing systems (kanji, hiragana, katakana), mixed script usage, and the lack of word boundaries. It was trained on large-scale Japanese text data to capture the nuances of the language effectively.

PLaMo Embedding 1B JP is recommended for Japanese-language search engines, RAG systems, and content recommendation applications requiring semantically aware text matching.