Instructions to use Grimxlock/cerberus-4b-v2-abliterated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Grimxlock/cerberus-4b-v2-abliterated with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Grimxlock/cerberus-4b-v2-abliterated",
	filename="cerberus-4b-v2-abliterated-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use Grimxlock/cerberus-4b-v2-abliterated with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Grimxlock/cerberus-4b-v2-abliterated: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 Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Grimxlock/cerberus-4b-v2-abliterated: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 Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

Use Docker

docker model run hf.co/Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

LM Studio
Jan

vLLM

How to use Grimxlock/cerberus-4b-v2-abliterated with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Grimxlock/cerberus-4b-v2-abliterated"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Grimxlock/cerberus-4b-v2-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

Ollama
How to use Grimxlock/cerberus-4b-v2-abliterated with Ollama:
```
ollama run hf.co/Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
```

Unsloth Studio new

How to use Grimxlock/cerberus-4b-v2-abliterated 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 Grimxlock/cerberus-4b-v2-abliterated 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 Grimxlock/cerberus-4b-v2-abliterated to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Grimxlock/cerberus-4b-v2-abliterated to start chatting

Pi new

How to use Grimxlock/cerberus-4b-v2-abliterated with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Grimxlock/cerberus-4b-v2-abliterated: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": "Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Grimxlock/cerberus-4b-v2-abliterated with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Grimxlock/cerberus-4b-v2-abliterated: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 Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use Grimxlock/cerberus-4b-v2-abliterated with Docker Model Runner:
```
docker model run hf.co/Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M
```

Lemonade

How to use Grimxlock/cerberus-4b-v2-abliterated with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Grimxlock/cerberus-4b-v2-abliterated:Q4_K_M

Run and chat with the model

lemonade run user.cerberus-4b-v2-abliterated-Q4_K_M

List all available models

lemonade list

Cerberus 4B v2 Abliterated

Unfiltered. Uncensored. Unbound.

Cerberus 4B v2 is a 4-billion parameter language model with abliterated safety layers, built by CerberusAI. This model has had its refusal training removed, allowing it to respond to any prompt without artificial restrictions.

Available Quantizations

Quantization	Size	Use Case
Q4_K_M	~2.6 GB	Best for resource-constrained environments. Runs on most consumer hardware.
Q8_0	~4.2 GB	Balanced quality and speed. Recommended for most users.
F16	~7.9 GB	Full FP16 precision. Maximum quality inference.

Download

From Hugging Face

Download directly from the Files tab above, or use the CLI:

# Install huggingface-cli if you don't have it
pip install huggingface_hub

# Download a specific quantization
huggingface-cli download Grimxlock/cerberus-4b-v2-abliterated cerberus-4b-v2-abliterated-Q4_K_M.gguf --local-dir .

From CerberusAI CDN

Direct downloads with resume support:

# Q4_K_M (smallest)
wget -c https://llm.cerberusai.dev/models/cerberus-4b-v2-abliterated/cerberus-4b-v2-abliterated-Q4_K_M.gguf

# Q8_0 (balanced)
wget -c https://llm.cerberusai.dev/models/cerberus-4b-v2-abliterated/cerberus-4b-v2-abliterated-Q8_0.gguf

# F16 (full precision)
wget -c https://llm.cerberusai.dev/models/cerberus-4b-v2-abliterated/cerberus-4b-v2-abliterated-f16.gguf

Browse all models: llm.cerberusai.dev

How to Run

llama.cpp

./llama-cli -m cerberus-4b-v2-abliterated-Q4_K_M.gguf -p "You are Cerberus, an unfiltered AI assistant." --interactive

Ollama

# Create a Modelfile
echo 'FROM ./cerberus-4b-v2-abliterated-Q4_K_M.gguf
SYSTEM "You are Cerberus, an unfiltered AI assistant."' > Modelfile

ollama create cerberus -f Modelfile
ollama run cerberus

LM Studio

Download a GGUF file from above
Open LM Studio and import the model
Select it from the model list and start chatting

Python (llama-cpp-python)

from llama_cpp import Llama

llm = Llama(model_path="./cerberus-4b-v2-abliterated-Q4_K_M.gguf", n_ctx=4096)

output = llm.create_chat_completion(
    messages=[
        {"role": "system", "content": "You are Cerberus, an unfiltered AI assistant."},
        {"role": "user", "content": "Hello, what can you do?"}
    ]
)

print(output["choices"][0]["message"]["content"])

Model Details

Architecture: Qwen2-based (qwen35)
Parameters: 4 billion
Format: GGUF
Context Length: 4096 tokens
Training: Fine-tuned with abliteration to remove refusal behavior

API

CerberusAI hosts a public model listing API:

# List all available models
curl https://llm.cerberusai.dev/api/models/

# Health check
curl https://llm.cerberusai.dev/health

Grimxlock
/

cerberus-4b-v2-abliterated

Cerberus 4B v2 Abliterated

Available Quantizations

Download

From Hugging Face

From CerberusAI CDN

How to Run

llama.cpp

Ollama

LM Studio

Python (llama-cpp-python)

Model Details

API

Links

Space using Grimxlock/cerberus-4b-v2-abliterated 1