Instructions to use KavinduHansaka/granite-4.0-h-tiny-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use KavinduHansaka/granite-4.0-h-tiny-gguf with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="KavinduHansaka/granite-4.0-h-tiny-gguf",
	filename="granite-4.0-h-tiny-Q4_0.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use KavinduHansaka/granite-4.0-h-tiny-gguf with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
# Run inference directly in the terminal:
llama-cli -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
# Run inference directly in the terminal:
llama-cli -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

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 KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
# Run inference directly in the terminal:
./llama-cli -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

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 KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

Use Docker

docker model run hf.co/KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

LM Studio
Jan

vLLM

How to use KavinduHansaka/granite-4.0-h-tiny-gguf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "KavinduHansaka/granite-4.0-h-tiny-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": "KavinduHansaka/granite-4.0-h-tiny-gguf",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

Ollama
How to use KavinduHansaka/granite-4.0-h-tiny-gguf with Ollama:
```
ollama run hf.co/KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
```

Unsloth Studio

How to use KavinduHansaka/granite-4.0-h-tiny-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 KavinduHansaka/granite-4.0-h-tiny-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 KavinduHansaka/granite-4.0-h-tiny-gguf to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for KavinduHansaka/granite-4.0-h-tiny-gguf to start chatting

How to use KavinduHansaka/granite-4.0-h-tiny-gguf with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

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": "KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use KavinduHansaka/granite-4.0-h-tiny-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 KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

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 KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use KavinduHansaka/granite-4.0-h-tiny-gguf with Docker Model Runner:
```
docker model run hf.co/KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0
```

Lemonade

How to use KavinduHansaka/granite-4.0-h-tiny-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull KavinduHansaka/granite-4.0-h-tiny-gguf:Q4_0

Run and chat with the model

lemonade run user.granite-4.0-h-tiny-gguf-Q4_0

List all available models

lemonade list

granite-4.0-h-tiny-gguf

🚀 GGUF-quantized release of IBM Granite 4.0 H Tiny, converted for efficient local inference.

This repository provides GGUF-format versions of the original
ibm-granite/granite-4.0-h-tiny model, enabling fast execution with llama.cpp, Ollama, LM Studio, and other GGUF-compatible runtimes.

Model Overview

Base Model: IBM Granite 4.0 H Tiny
Model Type: Decoder-only Transformer
Format: GGUF
Pipeline: Text Generation
Language: English

Designed for lightweight, low-latency inference while preserving strong instruction-following capabilities.

What This Repository Contains

✅ GGUF-converted model weights
✅ Quantized for CPU / low-VRAM environments
✅ Ready for local & edge inference
❌ No fine-tuning or weight modification (format conversion only)

Usage

Ollama

Create a Modelfile:

FROM granite-4.0-h-tiny.gguf

Run:

ollama create granite-tiny -f Modelfile
ollama run granite-tiny

LM Studio / GUI Tools

Open LM Studio
Load the .gguf file
Select a llama.cpp backend
Start inference

Quantization Notes

GGUF quantization provides:

Reduced memory usage
Faster load times
Compatibility with CPU-only systems

Recommended usage:

Q4 / Q5: Best balance of speed & quality
Q8: Higher quality, more memory

Intended Use Cases

Text generation
Review & sentiment analysis
QA automation pipelines
Agentic systems (RAG, MCP, LangGraph, CrewAI)
Offline / embedded AI applications

Attribution

Original Model:
IBM Granite Team
https://huggingface.co/ibm-granite/granite-4.0-h-tiny
GGUF Conversion & Packaging:
Kavindu Hansaka Jayasinghe

License

This repository follows the MIT license, consistent with the original model.

Please review the base model license before commercial usage.

Disclaimer

This is an unofficial GGUF conversion.
IBM is not affiliated with or responsible for this release.

Downloads last month: 1

GGUF

Model size

7B params

Architecture

granitehybrid

Hardware compatibility

4-bit

Model tree for KavinduHansaka/granite-4.0-h-tiny-gguf

Base model

ibm-granite/granite-4.0-h-tiny

Quantized

(33)

this model