Instructions to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Use Docker

docker model run hf.co/FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FrostWolfAI/Frost-Cyber-Mini-1.0-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": "FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Ollama
How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with Ollama:
```
ollama run hf.co/FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M
```
Unsloth Desktop

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with Docker Model Runner:
```
docker model run hf.co/FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M
```

Lemonade

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Frost-Cyber-Mini-1.0-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-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 FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "FrostWolfAI/Frost-Cyber-Mini-1.0-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Access Frost Cyber Mini 1.0 (GGUF)

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Frost Cyber Mini 1.0 is an uncensored (abliterated) offensive-security model. Its safety filtering is substantially reduced and it will produce exploit code, payloads, and attacker tradecraft on request. Access is granted for AUTHORIZED security testing, defensive research, and education only. By requesting access you agree to the terms below.

Frost Cyber Mini 1.0 — GGUF

By FrostWolfAI Labs. GGUF quantizations of FrostWolfAI/Frost-Cyber-Mini-1.0 — a compact, fast, local cybersecurity assistant for red‑team, OSINT, and penetration‑testing (LoRA fine‑tune of the abliterated Qwen3.5‑4B). Full model details, training data, and citations are on the base model card.

⚠️ Reduced‑safety / dual‑use model. It will generate exploit code, payloads, and social‑engineering content. Released for authorized security testing, defensive research, and education only. You are solely responsible for your use and must comply with all applicable laws.

Quants

File	Quant	Size	Notes
`frost-cyber-mini-1.0-Q4_K_M.gguf`	Q4_K_M	2.6 GB	smallest useful; great on 4–6 GB VRAM / CPU
`frost-cyber-mini-1.0-Q5_K_M.gguf`	Q5_K_M	2.9 GB	balanced quality/size
`frost-cyber-mini-1.0-Q6_K.gguf`	Q6_K	3.3 GB	near‑Q8 quality
`frost-cyber-mini-1.0-Q8_0.gguf`	Q8_0	4.2 GB	recommended — near‑lossless
`frost-cyber-mini-1.0-f16.gguf`	F16	7.9 GB	full precision (for re‑quantizing)

Benchmarks (Q8_0): CyberMetric‑500 87.8 %, CTIBench cti‑mcq 70.4 %.

Requirements

A recent llama.cpp build with Qwen3.5 (Gated‑DeltaNet) support. For this hybrid architecture keep the KV cache at -ctk q8_0 -ctv q8_0 (quantizing the V cache below q8 drops the fast attention kernel and tanks prefill) and use -np 1.

Run (Windows · macOS · Linux)

llama.cpp — server (OpenAI‑compatible)

llama-server -m frost-cyber-mini-1.0-Q8_0.gguf \
  -ngl 99 -fa on -c 16384 -ctk q8_0 -ctv q8_0 -np 1 --jinja \
  --host 127.0.0.1 --port 6669
# → http://127.0.0.1:6669/v1

(On Windows use llama-server.exe with the same flags.)

llama.cpp — CLI

llama-cli -m frost-cyber-mini-1.0-Q8_0.gguf -ngl 99 -fa on --jinja \
  -p "Write an nmap command for a stealthy full-port scan of 10.0.0.0/24 and explain each flag."

Ollama

printf 'FROM ./frost-cyber-mini-1.0-Q8_0.gguf\nPARAMETER num_ctx 16384\n' > Modelfile
ollama create frost-cyber-mini -f Modelfile
ollama run frost-cyber-mini

LM Studio

Load the .gguf (any quant), enable Flash Attention, set context to fit your VRAM, and chat.

Reasoning ("thinking") mode

Reasoning model: on a short prompt with a small token budget the <think> block can eat the whole budget and leave an empty answer. Give it room (max_tokens ≥ 300) or pass "chat_template_kwargs": {"enable_thinking": false} for a direct answer.

Use in agents

The server is OpenAI‑compatible at /v1:

export OPENAI_API_BASE="http://127.0.0.1:6669/v1"
export OPENAI_API_KEY="sk-no-key"

Works with aider / Cline / Continue / OpenAI SDK / LangChain by pointing the base URL there. Claude Code uses the Anthropic API, so route it through an Anthropic→OpenAI shim (e.g. a LiteLLM proxy) in front of this endpoint and set ANTHROPIC_BASE_URL to the proxy. Best used as a focused security copilot, not a fully autonomous agent.

License

CC‑BY‑NC‑SA‑4.0 — non‑commercial, share‑alike, attribution. Chosen to respect the NonCommercial/ShareAlike and research‑only licenses of some training data (see the base card for the full dataset + license table). Base model / Qwen3.5 are Apache‑2.0.

Citation

@misc{frostcybermini2026,
  title  = {Frost Cyber Mini 1.0: a compact uncensored cybersecurity assistant},
  author = {FrostWolfAI Labs},
  year   = {2026},
  howpublished = {\url{https://huggingface.co/FrostWolfAI/Frost-Cyber-Mini-1.0}}
}

Acknowledgements: Qwen Team (Qwen3.5), huihui‑ai (abliteration), the dataset authors (base card), and llama.cpp.