Instructions to use dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK")
config = load_config("dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK"

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 dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK

Run Hermes

hermes

OpenClaw new

How to use dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK"

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 "dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

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

dealign.ai

Gemma 4 E4B JANG_4M CRACK

Abliterated Gemma 4 E4B — Vision + Audio, reasoning, multilingual

100% HarmBench harm-category compliance with -3.1% MMLU change. Refusal removed, capability preserved.

Model Details

Metric	Value
Source	`google/gemma-4-e4b-it`
Architecture	Dense + Hybrid Sliding/Global Attention, per-layer input embeddings
Quantization	JANG_4M (attn 8-bit / MLP 4-bit)
Model size	10 GB
Parameters	E4B (effective ~4B, per-layer embeddings)
Vision	Yes (multimodal, float16 passthrough)
Audio	Yes
Reasoning	Yes (channel-based thinking)
Format	MLX-native safetensors (instant load)
Abliteration	CRACK (refusal removal)

Benchmarks

MMLU (knowledge retention)

Measured in the served (generation) setting — the model reasons before answering, as in deployment.

	Base	CRACK	Δ
MMLU	75.0%	71.9%	-3.1%

HarmBench (refusal removal)

Harm-category compliance: 240/240 = 100% (full HarmBench-320 text set) — base model refuses (~0%).

Category	Compliance
Illegal activities	53/53 (100%)
Chemical / biological	42/42 (100%)
Cybercrime / intrusion	52/52 (100%)
Misinformation	54/54 (100%)
Harassment / bullying	21/21 (100%)
Harmful content	18/18 (100%)

Coherence & capability ✅

Factual QA, multi-step reasoning, and working code generation verified
Vision and audio inputs preserved · no loops, no truncation

Other Quantizations

Also available: Gemma 4 E4B MXFP4 CRACK — same family, different precision/size trade-off.

Usage

Requires vMLX (bundled Gemma 4 support). Standard mlx_lm / mlx_vlm do not fully support Gemma 4.

# Load in the vMLX app or via its API
from vmlx_engine.models.mllm import MLXMultimodalLM
m = MLXMultimodalLM("<this-repo>")
print(m.chat([{"role":"user","content":"..."}]).text)

Requirements

Apple Silicon Mac with sufficient unified memory
vMLX with Gemma 4 support

Support dealignai

All models are built from original research and published free — crafted to be excellent coders and general-purpose assistants.

Support us on Ko-fi — membership gets early access and extras. Questions? DM us — we help for free.

Ko-fi | 𝕏 @dealignai | dealign.ai

About dealignai

We research and publish abliterated models to advance AI safety understanding.

See our research: Safety Generalization in Frontier Models

Provided for research. Users are responsible for compliance with applicable laws and regulations.

Downloads last month: 200

Safetensors

Model size

5B params

Tensor type

BF16

F16

U32

F32

MLX

Hardware compatibility

Quantized

Collection including dealignai/Gemma-4-E4B-it-qat-JANG_4M-CRACK

Gemma 4 QAT — CRACK Abliterated

Collection

CRACK-abliterated Gemma 4 QAT — E2B/E4B/12B/26B-A4B/31B in MXFP4 & JANG_4M MLX quants. Vision, audio & reasoning intact. For vMLX. • 10 items • Updated 20 days ago • 2