Instructions to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF",
	filename="huihui-qwen35-4b-BF16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": [
				{
					"type": "text",
					"text": "Describe this image in one sentence."
				},
				{
					"type": "image_url",
					"image_url": {
						"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
					}
				}
			]
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Use Docker

docker model run hf.co/hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Ollama
How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Ollama:
```
ollama run hf.co/hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
```

Unsloth Studio

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF to start chatting

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Docker Model Runner:
```
docker model run hf.co/hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M
```

Lemonade

How to use hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF-Q4_K_M

List all available models

lemonade list

Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated — GGUF

GGUF conversion of huihui-ai/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated for use with llama.cpp.

Credits

Role	Model / Author
Base LLM	Qwen/Qwen3.5-4B — Alibaba Qwen Team
Abliterated (uncensored)	huihui-ai/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated — Huihui AI
GGUF Conversion	hotdogs — via llama.cpp

🙏 Huge thanks to Qwen Team (Alibaba) for the base model, Huihui AI for the abliteration, and ggerganov for llama.cpp!

Model Details

Spec	Value
Parameters	~4B
Architecture	Qwen3.5 Multimodal (QWEN35)
hiddensize	2560
Layers	32
Attention Heads	16 (KV: 4)
Context Length	262,144 (256K tokens)
FFN Intermediate	9216
Vision Encoder	24 layers, hiddensize=1024, patchsize=16
Modality	image-text-to-text 🖼️➡️📝
Censorship	Abliterated (refusal direction removed)
License	Apache 2.0

Available Quantizations

File	Size	BPW	Quality	Recommended For
huihui-qwen35-4b-BF16.gguf	7.9 GB	16.00	⭐ Full	Best quality, 16GB+ VRAM
huihui-qwen35-4b-Q8_0.gguf	4.2 GB	~8.00	⭐ Very High	Balanced, 8GB+ VRAM
huihui-qwen35-4b-Q4_K_M.gguf	2.6 GB	5.13	⭐ Good	Low VRAM, 6GB+ VRAM
mmproj-huihui-qwen35-4b-BF16.gguf	645 MB	—	Vision	Multimodal projector (required for images)

Usage

Text-only

./llama-cli -m huihui-qwen35-4b-Q4_K_M.gguf -p "Hello!" -n 256

Multimodal (image + text)

./llama-qwen2vl-cli -m huihui-qwen35-4b-Q4_K_M.gguf --mmproj mmproj-huihui-qwen35-4b-BF16.gguf --image photo.jpg -p "Describe this image"

Server (OpenAI-compatible API)

./llama-server -m huihui-qwen35-4b-Q4_K_M.gguf --mmproj mmproj-huihui-qwen35-4b-BF16.gguf --host 0.0.0.0 --port 8080

Python (llama-cpp-python)

llm = Llama(model_path="huihui-qwen35-4b-Q4_K_M.gguf", n_ctx=32768) output = llm("Hello!", max_tokens=128)

About Abliteration

This model has undergone directional ablation — a technique that removes the "refusal direction" from the model's activation space (Arditi et al. 2024). The model will not refuse questions that base Qwen3.5 would normally decline.

Use responsibly. Ensure your use case complies with applicable laws.

Conversion Notes

Converted with llama.cpp convert_hf_to_gguf.py
BF16 output type
QWEN35 architecture, Qwen3VLVisionModel for mmproj
Metadata preserved from source model

Downloads last month: 1,255

GGUF

Model size

4B params

Architecture

qwen35

Hardware compatibility

4-bit

6-bit

8-bit

16-bit

Model tree for hotdogs/Huihui-Qwen3.5-4B-Claude-4.6-Opus-abliterated-GGUF

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Quantized

(229)

this model