Nous Releases Arete-27B, a Fine-Tune of Qwen3.8-27B
AI · First reported 6 Nov, 21:48 · Updated 22:16
Nous Research says its Arete-27B model reduced state-aligned responses on its evaluation while retaining similar results on one agentic benchmark. Its release includes model weights, evaluation prompts and training data.
- Arete-27B is a supervised fine-tune of Alibaba’s Qwen3.8-27B, trained on about 52.8 million tokens; Nous reports no reinforcement learning or architecture changes.
- On Nous’s 803-prompt evaluation, the share of answers classified as adopting state framing fell from 22% for Qwen3.8-27B to 0.9% for Arete-27B.
- Nous reports that fact scores improved but remained imperfect, while Hermes Bench results were 65.60% for Arete-27B and 65.77% for the base model.
Summary written by Lux AI from 3 articles
Key facts
- Model
- Arete-27B, a supervised fine-tune of Qwen3.8-27B.
- Evaluation size
- 803 prompts across seven categories.
- Reported state-framing rate
- 22% for Qwen3.8-27B; 0.9% for Arete-27B.
- Fact mean
- 2.32 for Qwen3.8-27B; 2.98 out of 5 for Arete-27B.
- Training data
- About 52.8 million tokens, according to Nous.
Source spectrum
Far left
Left
Lean left
Centre
Lean right
Right
Far right
Unrated
Left 0 · Centre 1 · Right 0 of 1 rated sources · +1 unrated
All coverage
Coverage over time
LeftCentreRight
Articles per hour since first reported, by side.
Ownership
- Unknown 50% (1)
- Corporation 50% (1)
Share of the 2 sources by owner type.