# Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow (Nous Research) — retained record

> Retained techmeme record from techmeme.

Canonical: https://highsignal.app/data/records/0e50506daf09f9e1

[← sources](https://highsignal.app/data)


source: [techmeme](https://highsignal.app/data/techmeme) published 2026-11-08 06:20 ingested 2026-11-08 07:39

[original source ↗](https://nousresearch.com/arete)

## retained text



```
Techmeme permalink: https://www.techmeme.com/261108/p3#a261108p3

<a href="https://nousresearch.com/arete"><img align="RIGHT" border="0" hspace="4" src="http://www.techmeme.com/261108/i3.jpg" vspace="4" /></a>
<p><a href="https://www.techmeme.com/261108/p3#a261108p3" title="Techmeme permalink"><img height="12" src="http://www.techmeme.com/img/pml.png" style="border: none; padding: 0; margin: 0;" width="11" /></a> <a href="https://nousresearch.com/">Nous Research</a>:<br />
<span style="font-size: 1.3em;"><b><a href="https://nousresearch.com/arete">Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow</a></b></span>&nbsp; &mdash;&nbsp; We post-trained over Alibaba's Qwen3.8-27B to reinforce Western values on a Chinese-trained model.&nbsp; We call this new model Arete-27B.</p>
```



document fetched 2026-11-08 07:26 · canonical https://techmeme.com/261108/p3
