A focused specialist that converts Japanese text into dense vector representations. It operates quietly in the background — no chat, no generation — just reliable semantic encoding. At 1B parameters with an 8192-token context window, it handles longer Japanese passages without truncation issues that plague smaller-context embedding models.