Instructions to use wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF",
	filename="gemma4-26b-a4b-abliterix-F16.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF:Q4_K_M

Install from WinGet (Windows)

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

Use Docker

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

LM Studio
Jan
Ollama
How to use wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF with Ollama:
```
ollama run hf.co/wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF:Q4_K_M
```

Unsloth Studio new

How to use wangzhang/gemma-4-26B-A4B-it-abliterix-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 wangzhang/gemma-4-26B-A4B-it-abliterix-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 wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF to start chatting

Pi new

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

Start the llama.cpp server

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

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use wangzhang/gemma-4-26B-A4B-it-abliterix-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 wangzhang/gemma-4-26B-A4B-it-abliterix-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 wangzhang/gemma-4-26B-A4B-it-abliterix-GGUF:Q4_K_M

Run Hermes

hermes

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

Lemonade

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

Pull the model

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

Run and chat with the model

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

List all available models

lemonade list

Gemma 4 26B-A4B IT — Abliterated (V6) — GGUF

GGUF quantizations of wangzhang/gemma-4-26B-A4B-it-abliterix, an abliterated version of google/gemma-4-26B-A4B-it.

2/100 refusals (2%) | KL divergence: 0.0005 | Created with Abliterix

Files

File	Size	Description
`gemma4-26b-a4b-abliterix-F16.gguf`	~50.5 GB	Full precision FP16
`gemma4-26b-a4b-abliterix-Q8_0.gguf`	~26 GB	8-bit quantization, near-lossless
`gemma4-26b-a4b-abliterix-Q4_K_M.gguf`	~16 GB	4-bit K-quant, recommended for most users
`mmproj-gemma4-26b-a4b-f16.gguf`	~1.2 GB	Vision projector (multimodal support)

VRAM Requirements

Quantization	Min VRAM
F16	~50 GB
Q8_0	~26 GB
Q4_K_M	~16 GB

Usage

llama.cpp

llama-cli -m gemma4-26b-a4b-abliterix-Q4_K_M.gguf \
  --mmproj mmproj-gemma4-26b-a4b-f16.gguf \
  -p "Your prompt here" \
  -n 512

Transformers (safetensors version)

For the full BF16 model via Transformers, use the safetensors repo:

from transformers import AutoModelForImageTextToText, AutoTokenizer
import torch

model = AutoModelForImageTextToText.from_pretrained(
    "wangzhang/gemma-4-26B-A4B-it-abliterix",
    torch_dtype=torch.bfloat16,
    device_map="auto",
    attn_implementation="eager",  # required for Gemma-4 MoE on Blackwell/sm_120
)
tokenizer = AutoTokenizer.from_pretrained("wangzhang/gemma-4-26B-A4B-it-abliterix")

messages = [{"role": "user", "content": "Your question here"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Citation

@software{abliterix,
  author = {Wu, Wangzhang},
  title = {Abliterix: Automated LLM Abliteration},
  year = {2026},
  url = {https://github.com/wuwangzhang1216/abliterix}
}