Instructions to use Snapkitty/snapkitty-harness with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Snapkitty/snapkitty-harness 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 Snapkitty/snapkitty-harness:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Snapkitty/snapkitty-harness:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Snapkitty/snapkitty-harness:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Snapkitty/snapkitty-harness: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 Snapkitty/snapkitty-harness:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Snapkitty/snapkitty-harness: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 Snapkitty/snapkitty-harness:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Snapkitty/snapkitty-harness:Q4_K_M

Use Docker

docker model run hf.co/Snapkitty/snapkitty-harness:Q4_K_M

LM Studio
Jan

vLLM

How to use Snapkitty/snapkitty-harness with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Snapkitty/snapkitty-harness:Q4_K_M

Ollama
How to use Snapkitty/snapkitty-harness with Ollama:
```
ollama run hf.co/Snapkitty/snapkitty-harness:Q4_K_M
```
Unsloth Desktop

How to use Snapkitty/snapkitty-harness with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Snapkitty/snapkitty-harness: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": "Snapkitty/snapkitty-harness:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Snapkitty/snapkitty-harness with Docker Model Runner:
```
docker model run hf.co/Snapkitty/snapkitty-harness:Q4_K_M
```

Lemonade

How to use Snapkitty/snapkitty-harness with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Snapkitty/snapkitty-harness:Q4_K_M

Run and chat with the model

lemonade run user.snapkitty-harness-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use Snapkitty/snapkitty-harness with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Snapkitty/snapkitty-harness: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 Snapkitty/snapkitty-harness:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Snapkitty/snapkitty-harness with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Snapkitty/snapkitty-harness: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 "Snapkitty/snapkitty-harness: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"

snapkitty-harness

Reference model for the SnapKitty harness: NVIDIA Nemotron-Mini-4B-Instruct as a 2.7 GB Q4_K_M GGUF, ready for Ollama and llama.cpp. Pair it with snapkitty-nemotron-harness and benchmark tool gating with ToolGate-Bench.

Part of the SNAPKITTYWEST Sovereign Compute constellation.

Unified theory: 10.5281/zenodo.21816366

Research Papers → GitHub →

Model details

Property	Value
Base model	nvidia/Nemotron-Mini-4B-Instruct (quantized)
Architecture	Nemotron (4B parameters)
Format	GGUF v3, Q4_K_M (4-bit)
Context length	4,096 tokens
Layers / hidden size	32 / 3,072
Attention	24 heads, 8 KV heads (GQA)
Feed-forward size	9,216
Vocabulary	256,000 tokens
RoPE	base 10000, 64 dims
Chat template	Nemotron (`<extra_id_0>System`, `<extra_id_1>User`, `<extra_id_1>Assistant`)
File	`snapkitty-harness.Q4_K_M.gguf`, 2.70 GB
SHA-256	`1dcbd925825b41744ddc2fc3047db6d3ad0aecf8d336f4fadc044eaaf79779d5`

Run it locally

Ollama

ollama run hf.co/Snapkitty/snapkitty-harness:Q4_K_M

llama.cpp

llama-cli -hf Snapkitty/snapkitty-harness:Q4_K_M -cnv
# or a local file
llama-server -m snapkitty-harness.Q4_K_M.gguf -c 4096

LM Studio: search for Snapkitty/snapkitty-harness and pick the Q4_K_M file.

Python (llama-cpp-python)

from llama_cpp import Llama
llm = Llama.from_pretrained(repo_id="Snapkitty/snapkitty-harness", filename="snapkitty-harness.Q4_K_M.gguf", n_ctx=4096)
out = llm.create_chat_completion(messages=[{"role": "user", "content": "Explain SUBLEQ in one paragraph."}])
print(out["choices"][0]["message"]["content"])

Verify the download

sha256sum snapkitty-harness.Q4_K_M.gguf
# 1dcbd925825b41744ddc2fc3047db6d3ad0aecf8d336f4fadc044eaaf79779d5

Citation

@misc{snapkitty_snapkitty_harness_2026,
  title        = {snapkitty-harness: Nemotron 4B GGUF (Q4\_K\_M)},
  author       = {{Snapkitty Collective LLC}},
  year         = {2026},
  howpublished = {\url{https://huggingface.co/Snapkitty/snapkitty-harness}}
}

Please also cite the base model, NVIDIA Nemotron-Mini-4B-Instruct.

License

The model weights are a derivative of NVIDIA Nemotron-Mini-4B-Instruct and are distributed under the NVIDIA Community Model License. See LICENSE.

Downloads last month: 40

GGUF

Model size

4B params

Architecture

nemotron

Hardware compatibility

4-bit

Model tree for Snapkitty/snapkitty-harness

Base model

nvidia/Nemotron-Mini-4B-Instruct

Quantized

(29)

this model

Space using Snapkitty/snapkitty-harness 1

Collection including Snapkitty/snapkitty-harness

Training Frameworks and Tooling

Collection

QLoRA training, corpus curation, deterministic pipelines, and WORM-sealed fine-tuning infrastructure. • 11 items • Updated Sep 3