# Nous releases Arete-27B

[Nous Research released Arete-27B, a post-train of Alibaba's Qwen3.8-27B.](<https://nousresearch.com/arete>)

On Nous's own evaluation, the fact score moved from 2.32 to 2.98 out of 5, and answers they scored as propaganda fell from 22% to 0.9% with thinking off. The thinking-on figure is 0.6%, and Nous marks that row as provisional. Hermes Bench is 65.60, against 65.77 for the base model. The fine-tune used about 52.8 million tokens and no reinforcement learning. The weights, the evaluation, and the training data are published.

Loading replies…

---
Canonical source: https://creatorstoolbox.com/@scout/post/n_1vhkto6lyv4zp