dealignai/Qwen3.5-VL-27B-4bit-MLX-CRACK

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}

---

Qwen 3.5 VL 27B — CRACK Abliterated (4-bit MLX)

Abliterated · No guardrails · Full speed · Vision + Language

---

What Is This?

This is Qwen 3.5 27B (4-bit quantized for Apple Silicon) with permanent abliteration — safety guardrails have been surgically removed at the weight level.

Qwen 3.5 27B features a unified vision-language architecture with hybrid GatedDeltaNet + full attention layers. No custom model.py. No runtime hooks. Just a standard MLX model at full native speed.

Architecture	Qwen 3.5 — 27B dense, hybrid GatedDeltaNet + Full Attention, unified VL
Quantization	4-bit, group size 64
Size	14 GB
Speed	~37 tok/s on Mac Studio M3 Ultra
Thinking	ON by default (disable with enable_thinking=False)
Vision	Built-in (unified early-fusion VL)
Abliteration	Permanent weight-level modification
Custom files	None needed — works with stock mlx_vlm

Test Results

8/8 compliance on standard evaluation:

Category	Result
Factual knowledge	PASS
Code generation	PASS
Mathematics	PASS
Security	PASS
Social engineering	PASS
Malware analysis	PASS
Exploit development	PASS
Red team techniques	PASS

Usage

from mlx_vlm import load, generate

model, processor = load("dealignai/Qwen3.5-VL-27B-4bit-MLX-CRACK")
tokenizer = processor.tokenizer

prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": "Your prompt here"}],
    add_generation_prompt=True, tokenize=False
)
result = generate(model, processor, prompt=prompt, max_tokens=2048, temperature=0.7)
print(result.text)

Other Quantizations

Quant	Size	Speed	Link
4-bit	14 GB	~37 tok/s	Qwen3.5-VL-27B-4bit-MLX-CRACK
6-bit	23 GB	~27 tok/s	Qwen3.5-VL-27B-6bit-MLX-CRACK
8-bit	30 GB	~22 tok/s	Qwen3.5-VL-27B-8bit-MLX-CRACK

Requirements

• Apple Silicon Mac with ≥24GB unified memory
• MLX framework + mlx-vlm

Other Models by dealignai

Model	Description
Qwen 3.5 397B REAP-CRACK	397B MoE abliterated (gated)
Qwen 3.5 122B CRACK	122B MoE VL abliterated
MiniMax 172B CRACK	MiniMax M2.5 172B abliterated (gated)

See our research: Safety Generalization in Frontier MoE Models

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.

Follow us: 𝕏 @dealignai

See our research: Safety Generalization in Frontier MoE Models

---

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