JANGQ-AI/Qwen3.5-397B-A17B-JANG_1L

MLX Studio — the only app that natively supports JANG models with reasoning

---

397B intelligence on a 128 GB Mac. This model fits in 112 GB — the first 397B quantization that runs on M4 Pro/Max 128 GB machines. Uses reasoning mode for 86.5% MMLU accuracy.

LM Studio, Ollama, oMLX do NOT support JANG format. Use MLX Studio or pip install "jang[mlx]>=2.1.5".

---

Qwen3.5-397B-A17B — JANG_1L (2.1-bit, 8-bit attention) — Reasoning + VLM

JANG — Jang Adaptive N-bit Grading | The GGUF Equivalent for MLX

---

     

JANG is fully open-source. Quantization engine, research, and full commit history: github.com/jjang-ai/jangq. Created by Jinho Jang.

Key Features

• 86.5% MMLU (200 questions, reasoning mode) — 397B on 128 GB Macs
• 36.1 tok/s generation, 96 tok/s prefill
• 112 GB on disk, 110 GB GPU RAM (peak 120 GB)
• Reasoning mode: <think>...</think> step-by-step problem solving
• Vision (VLM): 333 vision tensors, 31.6 tok/s image processing
• bfloat16 compute: auto-detected for 512-expert models

Results: JANG_1L vs MLX 4-bit (200-question MMLU)

Per-subject comparison across all modes. Both JANG and MLX 4-bit tested with and without reasoning.

Subject	JANG No-Think	JANG Reasoning	MLX 4-bit No-Think	MLX 4-bit Reasoning
Abstract Algebra	8/20	10/20	10/20	17/20
Anatomy	17/20	19/20	18/20	19/20
Astronomy	20/20	20/20	19/20	19/20
College CS	17/20	18/20	15/20	18/20
College Physics	17/20	18/20	15/20	19/20
HS Biology	19/20	20/20	19/20	19/20
HS Chemistry	17/20	18/20	17/20	19/20
HS Mathematics	8/20	10/20	12/20	19/20
Logical Fallacies	20/20	20/20	19/20	20/20
World Religions	19/20	20/20	19/20	19/20
Total	162/200 (81.0%)	173/200 (86.5%)	163/200 (81.5%)	188/200 (94.0%)

Summary

	JANG_1L	JANG_2L	MLX 4-bit	MLX 2/3-bit
MMLU (no-think)	81.0%	79.5%	81.5%	NaN -- cannot run
MMLU (reasoning)	86.5%	92.0%	94.0%	NaN -- cannot run
Size	112 GB	187 GB	209 GB	N/A
GPU RAM	110 GB	184 GB	~210 GB	N/A
Speed	36.1 tok/s	36.0 tok/s	~36 tok/s	N/A
Fits 128 GB?	YES	No	No	N/A

JANG_1L is 97 GB smaller than MLX 4-bit and fits on 128 GB Macs where MLX 4-bit (209 GB) cannot run. MLX 2-bit and 3-bit produce NaN -- cannot run (float16 overflow on 512-expert models). JANG solves this with bfloat16.

Specs

Metric	Value
Source	Qwen3.5-397B-A17B
Architecture	Hybrid MoE + SSM (GatedDeltaNet + Full Attention)
Experts	512 per layer, top-10 active (17B active params)
Profile	JANG_1L (CRITICAL=8, IMPORTANT=8, COMPRESS=2)
Average bits	2.13 bpw
Disk size	112 GB
GPU RAM	110 GB (peak 120 GB)
Speed	36.1 tok/s generation, 96 tok/s prefill
Compute	bfloat16 (auto-detected)
VLM	333 vision tensors, 31.6 tok/s

Requirements

• Apple Silicon Mac with 128+ GB unified memory
• MLX Studio (recommended) or pip install "jang[mlx]>=2.1.5"

Quick Start

pip install "jang[mlx]>=2.1.5"

from jang_tools.loader import load_jang_model
from mlx_lm import generate

model, tokenizer = load_jang_model("JANGQ-AI/Qwen3.5-397B-A17B-JANG_1L")

# With reasoning
messages = [{"role": "user", "content": "Prove that sqrt(2) is irrational."}]
prompt = tokenizer.apply_chat_template(messages, tokenize=False,
    add_generation_prompt=True, enable_thinking=True)
result = generate(model, tokenizer, prompt=prompt, max_tokens=2048)

# Without reasoning (faster)
prompt = tokenizer.apply_chat_template(messages, tokenize=False,
    add_generation_prompt=True, enable_thinking=False)
result = generate(model, tokenizer, prompt=prompt, max_tokens=100)

VLM Usage

from jang_tools.loader import load_jang_vlm_model
from mlx_vlm import generate as vlm_generate

model, processor = load_jang_vlm_model("JANGQ-AI/Qwen3.5-397B-A17B-JANG_1L")
messages = [{"role": "user", "content": [
    {"type": "image"},
    {"type": "text", "text": "Describe this image."},
]}]
prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
result = vlm_generate(model, processor, prompt=prompt, image=["photo.jpg"], max_tokens=200)

Technical Notes

• bfloat16: 512-expert models overflow float16 (max 65,504) at the shared expert down_proj. JANG auto-detects and uses bfloat16 (max 3.4x10^38). Zero quality impact.
• JANG_1L profile: 8-bit for all attention + routers + embeddings, 2-bit for expert MLP. On MoE models, expert MLP is 97.9% of params — so 2-bit covers almost everything while critical components get maximum precision.

---

JANG — Created by Jinho Jang (eric@jangq.ai) · @dealignai
GitHub · PyPI · HuggingFace

한국어

JANG_1L은 Qwen3.5-397B를 128 GB Mac에서 실행할 수 있는 최초의 양자화입니다. 112 GB, 36 tok/s, 86.5% MMLU.

pip install "jang[mlx]>=2.1.5"