inclusionAI/TwinFlow-Z-Image-Turbo

<h1 align="center">TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows</h1>

<div align="center">

&#160; &#160; &#160; &#160; <a href="https://arxiv.org/abs/2512.05150" target="_blank"><img src="https://img.shields.io/badge/Paper-b5212f.svg?logo=arxiv" height="21px"></a>

</div>

News

• —We release experimental version of faster Z-Image-Turbo!
• —We release TwinFlow-Qwen-Image-v1.0! And we are also working on Z-Image-Turbo to make it more faster!

TwinFlow

Checkout 2-NFE visualization of TwinFlow-Z-Image-Turbo-exp 👇

<details> <summary>👀 Original Z-Image-Turbo 2-NFE</summary>

<div align="center"> <img src="https://raw.githubusercontent.com/inclusionAI/TwinFlow/refs/heads/main/assets/zturbo2step.jpg" width="1000" /> <p style="margin-top: 8px; font-size: 14px; color: #666; font-weight: bold;"> 2-NFE visualization of Z-Image-Turbo </p> </div>

</details>

Overview

We introduce TwinFlow, a framework that realizes high-quality 1-step and few-step generation without the pipeline bloat.

Instead of relying on external discriminators or frozen teachers, TwinFlow creates an internal "twin trajectory". By extending the time interval to $t\in[−1,1]$, we utilize the negative time branch to map noise to "fake" data, creating a self-adversarial signal directly within the model.

Then, the model can rectify itself by minimizing the difference of the velocity fields between real trajectory and fake trajectory, i.e. the $\Delta_\mathrm{v}$. The rectification performs distribution matching as velocity matching, which gradually transforms the model into a 1-step/few-step generator.

Key Advantages:

• —One-model Simplicity. We eliminate the need for any auxiliary networks. The model learns to rectify its own flow field, acting as the generator, fake/real score. No extra GPU memory is wasted on frozen teachers or discriminators during training.
• —Scalability on Large Models. TwinFlow is easy to scale on 20B full-parameter training due to the one-model simplicity. In contrast, methods like VSD, SiD, and DMD/DMD2 require maintaining three separate models for distillation, which not only significantly increases memory consumption—often leading OOM, but also introduces substantial complexity when scaling to large-scale training regimes.

Inference Demo

Install the latest diffusers:

pip install git+https://github.com/huggingface/diffusers

Run inference demo inference.py:

python inference.py

Citation

@article{cheng2025twinflow,
  title={TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows},
  author={Cheng, Zhenglin and Sun, Peng and Li, Jianguo and Lin, Tao},
  journal={arXiv preprint arXiv:2512.05150},
  year={2025}
}

Acknowledgement

TwinFlow is built upon RCGM and UCGM, with much support from InclusionAI.