Crazy that it's a small lab becoming the frontier in term of moderation models, instead of Meta which is pouring dozens of billions into LLMs.

Meta would really benefit from work done on this front, however their model Llama Guards are quite lagging compared to the competition.