Instructions to use unsloth/gemma-4-26B-A4B-it-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 unsloth/gemma-4-26B-A4B-it-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 unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-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 unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-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 unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

Use Docker

docker model run hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

LM Studio
Jan

vLLM

How to use unsloth/gemma-4-26B-A4B-it-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/gemma-4-26B-A4B-it-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": "unsloth/gemma-4-26B-A4B-it-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/unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

Ollama
How to use unsloth/gemma-4-26B-A4B-it-GGUF with Ollama:
```
ollama run hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
```
Unsloth Desktop

How to use unsloth/gemma-4-26B-A4B-it-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-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": "unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use unsloth/gemma-4-26B-A4B-it-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M
```

Lemonade

How to use unsloth/gemma-4-26B-A4B-it-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

Run and chat with the model

lemonade run user.gemma-4-26B-A4B-it-GGUF-UD-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use unsloth/gemma-4-26B-A4B-it-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 unsloth/gemma-4-26B-A4B-it-GGUF:UD-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 unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use unsloth/gemma-4-26B-A4B-it-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/gemma-4-26B-A4B-it-GGUF:UD-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 "unsloth/gemma-4-26B-A4B-it-GGUF:UD-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"

Insanely good 2-bit quant

#40

by cogitech - opened May 6

Discussion

cogitech

May 6

I decided to try UD-Q2_K_XL on my RTX3060 12GB with 64k KV cache (at Q4_0) just for the hell of it. I am running it via llama.cpp's llama-server which is serving Hermes Agent. It is blowing my mind! This 2-bit model successfully navigated a gauntlet of "AI traps" that usually break much larger models. Here is the summary of its wins:

The Einstein Riddle (Zebra Puzzle): Proved complex constraint satisfaction by correctly mapping 15+ variables (nationalities, houses, pets) without losing track of the "bottleneck" clues.

The Car Wash Challenge: Demonstrated spatial awareness and object permanence by recognizing that walking to a car wash doesn't move the car.

The 30 Shirts Task: Showed an understanding of parallel vs. serial processing, correctly identifying that more items don't increase time if the heat source (the sun) is constant.

Sally’s Sisters: Solved a Theory of Mind trap by accurately calculating family relationships and recognizing that Sally herself is one of the sisters.

The Killer in the Room: Handled state-change logic and self-identity, realizing that the user’s actions (killing) added a new killer to the total count.

The Boiling Water/Freezing Room: Applied thermodynamics and temporal reasoning, correctly predicting that 10 minutes is insufficient for thermal equilibrium.

This is currently the best way to use 12GB of VRAM, IMHO. Although, I do realize for pure coding prowess, Qwen3.6 is probably better. That isn't my primary use case, so this 2-bit quant of Gemma4 wins, and it does it all so eloquently that I actually enjoy chatting with it.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment