A Japanese-optimized large language model by NII, available in sizes from 150M to 172B. Trained on llm-jp-corpus v3, supports SFT and DPO via Instruct2/3, and is Apache 2.0 licensed for models up to 13B.
Parameters
Undisclosed
Context Window
License
Proprietary
Release Date
2025-05-23
API Pricing
API pricing for this model is not yet available
Strengths
- •Largest 13B model with advanced instruction tuning.
- •Fourth iteration suggests refined alignment.
- •High capability for complex Japanese tasks.
Weaknesses
- •Very high computational and memory requirements.
- •Inference speed is significantly slower.
- •Not suitable for consumer-grade hardware.
Use Cases
- •High-end research and enterprise applications.
- •Complex document analysis and summarization.
- •Advanced Japanese language understanding tasks.