7d ago · r/LocalLLaMA →
from NousResearch: Arete-27B We post-trained over Alibaba's Qwen3.8-27B to reinforce western values on a Chinese-trained model.