---
title: Qwen3.5-27B-heretic-8bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/Qwen3.5-27B-heretic-8bit"
md_url: "https://www.modelscope.cn/models/mlx-community/Qwen3.5-27B-heretic-8bit.md"
repository: mlx-community/Qwen3.5-27B-heretic-8bit
last_updated: 2026-03-02
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen3_5
architectures:
  - Qwen3_5ForConditionalGeneration
base_model:
  - coder3101/Qwen3.5-27B-heretic
base_model_relation: quantized
parameters: 8.0B
tensor_type:
  - BF16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 68
stars: 0
tags:
  - mlx
  - quantized
  - 8bit
  - vision
  - vlm
  - qwen3.5
  - abliterated
  - uncensored
---

# Qwen3.5-27B-heretic-8bit

> Qwen3.5-27B-heretic-8bit - mlx-community 在 ModelScope 开源的模型。mlx-community/Qwen3.5-27B-heretic-8bit

mlx-community/Qwen3.5-27B-heretic-8bit 是 ModelScope 魔搭社区上的 8.0B 参数image-text-to-text模型，采用 apache-2.0 许可，基于 coder3101/Qwen3.5-27B-heretic 构建。

- **Repository**: mlx-community/Qwen3.5-27B-heretic-8bit
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 8.0B
- **Base model**: coder3101/Qwen3.5-27B-heretic
- **Tags**: mlx, quantized, 8bit, vision, vlm, qwen3.5, abliterated, uncensored
- **Downloads**: 68
- **Stars**: 0
- **Last updated**: 2026-03-02

Source: https://www.modelscope.cn/models/mlx-community/Qwen3.5-27B-heretic-8bit

---

# mlx-community/Qwen3.5-27B-heretic-8bit

This model was converted to MLX format from [`coder3101/Qwen3.5-27B-heretic`](https://huggingface.co/coder3101/Qwen3.5-27B-heretic) using [mlx-vlm](https://github.com/Blaizzy/mlx-vlm) version **0.3.12**.

Qwen3.5-27B-heretic is a **decensored/abliterated** version of [Qwen/Qwen3.5-27B](https://huggingface.co/Qwen/Qwen3.5-27B), created using **Heretic v1.2.0** with **Magnitude-Preserving Orthogonal Ablation (MPOA)**.

## Quantization Details

- **Bits**: 8
- **Group size**: 64
- **Mode**: affine
- **Total size**: ~29.5 GB
- **Bits per weight**: 8.627 (avg)

Vision encoder layers with dimensions incompatible with the group size are kept in bfloat16.

## Key Features

- **Vision + Text**: Natively multimodal — accepts images and video as input
- **Abliterated**: Refusal rate reduced from 94% to 14% while maintaining low KL divergence (0.0653) from the original model
- **Hybrid Architecture**: Gated DeltaNet + Gated Attention layers (not standard transformer)
- **Long Context**: 262,144 token context window
- **Thinking Mode**: Supports `<think>` reasoning (disabled by default in this conversion's chat template for faster interactive use; pass `enable_thinking=True` to re-enable)

## Use with mlx-vlm

```bash
pip install mlx-vlm
```

### Text Generation

```python
from mlx_vlm import load, generate

model, processor = load("mlx-community/Qwen3.5-27B-heretic-8bit")

prompt = "Hey, what's up?"
output = generate(model, processor, prompt, max_tokens=500, temperature=0.7)
print(output)
```

### Vision (Image Understanding)

```python
from mlx_vlm import load, generate

model, processor = load("mlx-community/Qwen3.5-27B-heretic-8bit")

output = generate(
    model, processor,
    prompt="Describe this image in detail",
    image="path/to/image.jpg",
    max_tokens=500,
    temperature=0.3,
)
print(output)
```

### CLI

```bash
# Text chat
mlx_vlm generate --model mlx-community/Qwen3.5-27B-heretic-8bit \
    --prompt "Hello" --max-tokens 500 --chat

# Image + text
mlx_vlm generate --model mlx-community/Qwen3.5-27B-heretic-8bit \
    --image photo.jpg --prompt "What do you see?" --max-tokens 500
```

## Re-enabling Thinking Mode

The chat template defaults to thinking OFF for snappier interactive use. To re-enable `<think>` reasoning, pass `enable_thinking=True` when applying the chat template:

```python
prompt = processor.tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, enable_thinking=True
)
```

## Original Model

- **Base**: [Qwen/Qwen3.5-27B](https://huggingface.co/Qwen/Qwen3.5-27B)
- **Abliteration**: [coder3101/Qwen3.5-27B-heretic](https://huggingface.co/coder3101/Qwen3.5-27B-heretic) (Heretic v1.2.0, MPOA method)
- **Parameters**: 27B
- **Architecture**: `Qwen3_5ForConditionalGeneration` (Vision-Language, Gated DeltaNet + Gated Attention)