LLM-jp-3 13B
Japan's national open LLM: 13B trained on 2.1T tokens of Japanese-heavy data.
About
LLM-jp-3 13B is the flagship dense model of Japan's LLM-jp project, developed by NII's Research and Development Center for Large Language Models and released in September 2024. Pre-trained on 2.1 trillion tokens of the llm-jp-corpus v3, it addresses the gap left by global models that underperform on Japanese knowledge, culture, and language.
Released under Apache 2.0 with open datasets, fine-tuning code, and evaluation results, the project later added Instruct2/Instruct3 variants with DPO-based safety improvements and a 3.1 refresh. With over 2,000 participating researchers, it is one of the world's most transparent national LLM programs.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 90; Trending score 70/100 from researcher assessment.