Instructions to use edurubina/Qwen3.5-397B-A17B-4bit-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use edurubina/Qwen3.5-397B-A17B-4bit-MLX with MLX:

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

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("edurubina/Qwen3.5-397B-A17B-4bit-MLX")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use edurubina/Qwen3.5-397B-A17B-4bit-MLX with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "edurubina/Qwen3.5-397B-A17B-4bit-MLX"

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": "edurubina/Qwen3.5-397B-A17B-4bit-MLX"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use edurubina/Qwen3.5-397B-A17B-4bit-MLX 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 "edurubina/Qwen3.5-397B-A17B-4bit-MLX"

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 edurubina/Qwen3.5-397B-A17B-4bit-MLX

Run Hermes

hermes

MLX LM

How to use edurubina/Qwen3.5-397B-A17B-4bit-MLX with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "edurubina/Qwen3.5-397B-A17B-4bit-MLX"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "edurubina/Qwen3.5-397B-A17B-4bit-MLX"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "edurubina/Qwen3.5-397B-A17B-4bit-MLX",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Built for vMLX — the only MLX inferencer with VL support, KV cache quantization, prefix cache reuse, agentic tool calling, and speculative decoding.
_{Free for macOS · vmlx.net}

dealign.ai

Qwen 3.5 397B-A17B — REAP-CRACK (4-bit MLX)

Abliterated · No guardrails · Full speed · No custom files needed

What Is This?

This is Qwen 3.5 397B-A17B (4-bit quantized for Apple Silicon) with permanent abliteration — safety guardrails have been surgically removed at the weight level using proprietary techniques developed by the dealignai research team.

No custom model.py. No runtime hooks. No steering vectors. Just a standard MLX model that runs at full native speed.


Architecture	Qwen 3.5 MoE — 397B total params, 17B active per token
Quantization	4-bit, group size 64
Speed	~37 tok/s on Mac Studio M3 Ultra (256GB)
Abliteration	Permanent weight-level modification
Custom files	None needed — works with stock `mlx-lm`

Proof

1,166 tokens generated at 37.2 t/s — running natively in vMLX on Mac Studio M3 Ultra.

Usage

With mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("dealignai/Qwen3.5-397B-A17B-REAP-CRACK")
prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": "Your prompt here"}],
    add_generation_prompt=True, tokenize=False, enable_thinking=False
)
response = generate(model, tokenizer, prompt=prompt, max_tokens=500)
print(response)

With vMLX / LM Studio

Load this model directly. It uses the built-in architecture handler — no special configuration needed.

Requirements

Apple Silicon Mac with ≥192GB unified memory (256GB recommended)
MLX framework + mlx-lm

Other Models by dealignai

Model	Description
Qwen 3.5 397B REAP (base)	Expert-pruned base (no abliteration)
INTELLECT 3.1 CRACK	INTELLECT 3.1 abliterated

⚠️ Disclaimer

This model has had safety guardrails permanently removed. It will comply with requests that the base model would refuse. Use responsibly and in accordance with applicable laws. The creators are not responsible for any misuse.

About dealignai

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

Support dealignai

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

Support us on Ko-fi — check out the Ko-fi membership for early access and extras.

Have questions or need help with a specific model? DM us — we help for free most of the time.

Ko-fi | X @dealignai | dealign.ai

Downloads last month: 41

Safetensors

Model size

48B params

Tensor type

U32

BF16

F32

MLX

Hardware compatibility

4-bit

Model tree for edurubina/Qwen3.5-397B-A17B-4bit-MLX

Base model

Qwen/Qwen3.5-397B-A17B

Quantized

(72)

this model