G9v3-39A5B-4bits-mlx

G9v3-39A5B — MLX 4bit

Quantized to 4 bits per weight.

Converted from ai9stars/G9v3-39A5B using mlx-lm.

Requirements

The g9v3 model type is support in my mlx-lm fork(AlphaKure/mlx-lm).

PR is awating review on mlx-lm

Update at: 2026/09/09

pip install git+https://github.com/AlphaKure/mlx-lm.git@02dc6aa00f9a79cdedd5e48f5ca1a302b4bebcd5

Use

mlx_lm.generate --model AlphaOxO/G9v3-39A5B-4bits-mlx --prompt "hello"

from mlx_lm import load, generate

model, tokenizer = load("AlphaOxO/G9v3-39A5B-4bits-mlx")
prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": "hello"}], add_generation_prompt=True
)
print(generate(model, tokenizer, prompt=prompt))