---
title: "Huihui Strips Alibaba's 177B Qwen3.8-Flash-Next of All Refusals"
subtitle: "A 177B-parameter Qwen3.8-Flash-Next variant ships in GGUF form with refusal behavior surgically removed, ready to run locally through llama.cpp."
url: "https://alphasignal.ai/news/huihui-strips-alibaba-s-177b-qwen3-8-flash-next-of-all-refusals"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-09-06T04:21:12.000Z"
updated: "2026-09-30T00:57:01.734Z"
type: "Model"
topics:
  - "Llms"
  - "Security"
subtopics:
  - "Small Models"
  - "Red Teaming"
upvotes: 172
source_url: "https://huggingface.co/huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated-GGUF"
comments: 0
truncated: true
---

# Huihui Strips Alibaba's 177B Qwen3.8-Flash-Next of All Refusals

> A 177B-parameter Qwen3.8-Flash-Next variant ships in GGUF form with refusal behavior surgically removed, ready to run locally through llama.cpp.

# Huihui releases an abliterated Qwen3.8-Flash-Next GGUF

The [Huihui-ai profile](https://huggingface.co/huihui-ai) has published a weight-edited version of Alibaba’s Qwen3.8-Flash-Next model. The [GGUF repository](https://huggingface.co/huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated-GGUF) packages the model for local inference through llama.cpp and compatible applications, with reduced refusal behavior as its defining modification. The repository displayed roughly 166,000 downloads at publication, although Hugging Face download counts represent qualifying file requests rather than unique users or completed installations.

| Detail                | Repository claim                           |
| --------------------- | ------------------------------------------ |
| Base model            | Qwen3.8-Flash-Next                         |
| Architecture          | 177B-parameter qwen4exp mixture of experts |
| Modification          | Refusal-direction abliteration             |
| Distribution format   | GGUF quantizations                         |
| Featured quantization | UD-Q4\_K\_XL, approximately 111 GB         |
| Advertised context    | Up to 262,144 tokens                       |
| License               | qwen-community-1.0                         |

## How one vector becomes a weight edit

Abliteration uses an empirical finding from the [original study](https://arxiv.org/abs/2406.11717): refusal behavior in several instruction-tuned models correlates with a direction in the residual stream, the shared hidden state passed between transformer layers. Removing that direction can suppress many explicit refusals while retaining much of the model’s previous behavior.

The procedure generally follows four steps:

1. Run matched sets of refusal-triggering and benign prompts through the model.
2. Record hidden activations at selected layers and token positions.
3. Calculate candidate refusal directions from the difference between the two activation sets.
4. Project the selected direction out of matrices that write to the residual stream, including embeddings, attention output projections, and MLP output projections.

With a unit refusal vector `r_hat`, the conceptual projection is:

```
P = I - r_hat @ r_hat.T
h_projected = P @ h
```

The implementation applies the equivalent projection to each targeted weight matrix. The multiplication side depends on how that matrix stores its output dimension. This edit suppresses the measured linear feature; refusal behavior can also depend on other directions, prompt distributions, decoding settings, and post-training biases.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Huihui-Qwen3.8-Flash-Next-abliterated-GGUF model card](https://huggingface.co/huihui-ai/Huihui-Qwen3.8-Flash-Next-abliterated-GGUF)
- [Heretic vs Abliterated: Compare Uncensored LLM Methods](https://privatellm.app/blog/heretic-vs-abliterated-uncensored-llm-comparison)
- [Uncensor any LLM with abliteration](https://huggingface.co/blog/mlabonne/abliteration)
- [remove-refusals-with-transformers](https://github.com/Sumandora/remove-refusals-with-transformers)
- [Abliterated Models Guide 2026](https://locallyuncensored.com/blog/abliterated-models-guide.html)