---
license: other
license_name: qwen-research
license_link: https://huggingface.co/Qwen/Qwen-Image-2.1/blob/main/LICENSE
base_model: Qwen/Qwen-Image-2.1
base_model_relation: quantized
library_name: mflux
pipeline_tag: text-to-image
tags:
- mlx
- mflux
- qwen-image
- text-to-image
- apple-silicon
---

# Qwen-Image-2.1 · MLX 8-bit (mflux)

[Qwen/Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1) quantized to 8-bit for Apple Silicon with
[mflux](https://github.com/mflux-community/mflux) (`mflux-save -q 8`, commit `8c00dab2`).

| Component | Precision | Size |
|---|---|---|
| Transformer (7B single-stream DiT) | 8-bit (MLX affine, group 64) | 7.1 GiB |
| Text encoder (Qwen3-VL-8B) | bf16 (mflux keeps it unquantized) | 14 GiB |
| VAE (64-channel) | 8-bit linear layers, conv layers bf16 | 1.2 GiB |
| **Total** | | **~23 GB** |

## Usage

Requires an mflux build that includes Qwen-Image-2.1 support (mflux `main` at or after `8c00dab2`).

```sh
mflux-generate-qwen-2.1 \
  --model JoyFusionAI/Qwen-Image-2.1-MLX-8bit \
  --base-model qwen-image-2.1 \
  --prompt "A red fox in a snowy forest holding a wooden sign that says \"Hello\"" \
  --steps 40 --low-ram
```

`--low-ram` releases the text encoder after prompt encoding and tiles the VAE decode. Measured on an
M1 Max at 768×768: peak memory ~10 GB (vs. 33–36 GB without it), ~6.7 s/step, no visible quality change.

Defaults recommended by the model authors: 40 steps, no guidance. Image sizes must be divisible by 16.

## License

These weights are a derivative of Qwen-Image-2.1 and are distributed under the
**Qwen Research License Agreement**. Non-commercial research and evaluation use only; commercial use
requires a separate license from the Qwen team. See the [original license](https://huggingface.co/Qwen/Qwen-Image-2.1/blob/main/LICENSE).

Qwen-Image-2.1 is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) Alibaba Cloud. All Rights Reserved.