Nous Research releases Arete-27B, Qwen3.8-27B post-trained to cut a "party line" rate from 22% to 1%

Nous Research announced Arete-27B on X on November 6. It is Qwen3.8-27B post-trained, in the lab's words, "to reinforce Western values". Nous argues that the strongest open models are Chinese and ship with state-aligned behavior in their weights.

To measure this, Nous built an 803-prompt evaluation across seven value categories, including freedom of expression, democracy and privacy, and counted how often a model adopts China's party line. Nous says that after fine-tuning on roughly 53 million tokens, Arete-27B's rate fell from 22% to 1%, in line with the Western models it tested, and that its Hermes Bench score shows no significant performance tradeoff. These are Nous's own figures.

Nous says the evaluation, the weights and the training data are released, with a blog post at nousresearch.com/arete.