# Huihui-Qwen3.8-27B-abliterated

> One of the most downloaded abliterated Qwen 3.8 27B builds. Strong general performance with significantly reduced refusals.


- **Base model:** Qwen/Qwen3.8-27B
- **Family:** Qwen
- **Size:** 27B
- **Method:** abliteration
- **Approx. VRAM (Q4):** ~16–18 GB (Q4)
- **License:** Apache-2.0
- **Quantizations:** GGUF (Q2–Q8), Transformers, NVFP4
- **Last updated:** 2026-11-18
- **Hugging Face:** https://huggingface.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF
- **Ollama:** `ollama run huihui_ai/Qwen3.8-abliterated:27b`

## Notes

One of the most downloaded abliterated Qwen 3.8 27B builds. Strong general performance with significantly reduced refusals.

Canonical page: https://www.wahloon.net/models/huihui-qwen3.8-27b-abliterated

You are responsible for how you use this model.