CRITICAL FIX (2026-03-19): Fixed eos_token_id — previous versions caused infinite thinking loops. You MUST re-download this model if you downloaded before today.

Update (2026-03-18): Models have been updated to v2.1.0 with VLM support, proper tokenizer, and fixed configs. If you downloaded before this date, please re-download for full MLX Studio compatibility.

MLX Studio — the only app that natively supports JANG models

Early Adoption: LM Studio, Ollama, oMLX, Inferencer do not support JANG yet. Use MLX Studio or pip install "jang[mlx]". Ask your favorite app's creators to add JANG support!

JANG

Qwen3.5-9B — JANG_4S (4.34-bit) — VLM

JANG — Jang Adaptive N-bit Grading | Mixed-Precision Quantization for Apple Silicon

GitHub

X/Twitter

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

Results (200-question MMLU)

Model	MMLU	Size	Speed
JANG_4S (4.34-bit)	73.0%	6.0 GB	—
MLX 4-bit	72.5%	4.7 GB	—
MLX 3-bit	64.0%	3.7 GB	—
MLX 2-bit	22.0%	2.6 GB	—

JANG_4S beats MLX 4-bit on 9B — attention at 6-bit preserves quality.

Per-Subject Scores

Subject	JANG_4S	MLX_4bit	MLX_3bit	MLX_2bit
Abstract Algebra	9/20	11/20	8/20	4/20
Anatomy	16/20	15/20	13/20	6/20
Astronomy	20/20	20/20	16/20	5/20
College CS	14/20	13/20	10/20	7/20
College Physics	13/20	13/20	12/20	6/20
HS Biology	18/20	18/20	19/20	4/20
HS Chemistry	15/20	14/20	15/20	4/20
HS Mathematics	8/20	9/20	5/20	2/20
Logical Fallacies	17/20	16/20	16/20	3/20
World Religions	16/20	16/20	14/20	3/20
Total (/200)	146	145	128	44

Specs

Metric	Value
Source	Qwen3.5-9B
Profile	JANG_4S (CRITICAL=6, IMPORTANT=4, COMPRESS=4)
Average bits	4.34
VLM	Yes (333 vision tensors)
Speed	~70 tok/s
Format	v2 (MLX-native, instant load)

Install

pip install "jang[mlx]"

For Vision-Language models:

pip install "jang[vlm]"

Quick Start

python

from jang_tools.loader import load_jang_model
from mlx_lm.sample_utils import make_sampler
from mlx_lm.generate import generate_step
import mlx.core as mx

model, tokenizer = load_jang_model("JANGQ-AI/Qwen3.5-9B-JANG_4S")
sampler = make_sampler(temp=0.7)

tokens = tokenizer.encode("What is photosynthesis?")
for tok, _ in generate_step(prompt=mx.array(tokens), model=model, max_tokens=200, sampler=sampler):
    t = tok.item() if hasattr(tok, 'item') else int(tok)
    print(tokenizer.decode([t]), end="", flush=True)
    if t == tokenizer.eos_token_id:
        break

VLM Inference

python

from jang_tools.loader import load_jang_vlm_model
from mlx_vlm import generate

model, processor = load_jang_vlm_model("JANGQ-AI/Qwen3.5-9B-JANG_4S")

prompt = processor.tokenizer.apply_chat_template(
    [{"role": "user", "content": [
        {"type": "image", "image": "photo.jpg"},
        {"type": "text", "text": "Describe this image."}
    ]}], add_generation_prompt=True, tokenize=False, enable_thinking=False)

result = generate(model, processor, prompt, ["photo.jpg"], max_tokens=200)
print(result.text)

한국어

Qwen3.5-9B — JANG 4S

JANG은 Apple Silicon을 위한 혼합정밀도 양자화 포맷입니다. MLX를 위한 GGUF와 같은 역할을 합니다.

모델	MMLU	크기
JANG_4S	73.0%	6.0 GB
MLX 4-bit	72.5%	4.7 GB

설치

pip install "jang[mlx]"

호환성

현재 **MLX Studio**만 JANG 포맷을 기본 지원합니다. LM Studio, Ollama 등은 아직 지원하지 않습니다.

GitHub · HuggingFace · MLX Studio · PyPI

장진호 제작 · Created by Jinho Jang — jangq.ai · @dealignai

JANGQ-AI/Qwen3.5-9B-JANG_4S