llm-jp/awesome-japanese-llm

The project 'awesome-japanese-llm' is a highly regarded curated list (1,370+ stars) maintained by LLM-jp, a significant Japanese research consortium. Its defensibility is not technical, but social and informational; it acts as a Schelling point for the Japanese AI community. While the 'code' is trivial to clone, the institutional backing of LLM-jp (NII and partners) gives it authoritative weight that a random fork would lack. Frontier labs like OpenAI or Google represent low risk here because they are model providers, not ecosystem archivists. The primary threat comes from automated tracking platforms like Hugging Face's Leaderboard, which could provide more dynamic, data-driven comparisons of Japanese models, potentially making manual lists obsolete. The 1,000+ day age and steady star growth indicate it is the de facto reference for the local market, but its moat is shallow and relies entirely on the maintainers' continued manual effort to stay current in a rapidly moving field.