Step1X-Edit: a unified image editing model performs impressively on various genuine user instructions.
## 1. Introduction
We release a state-of-the-art image editing model, **Step1X-Edit**, which can provide comparable performance against the closed-source models like GPT-4o and Gemini2 Flash. More specifically, we adopt the Multimodal LLM to process the reference image and user's editing instruction. A latent embedding has been extracted and integrated with a diffusion image decoder to obtain the target image. To train the model, we build a data generation pipeline to produce a high-quality dataset. For evaluation, we develop the GEdit-Bench, a novel benchmark rooted in real-world user instructions. Experimental results on GEdit-Bench demonstrate that Step1X-Edit outperforms existing open-source baselines by a substantial margin and approaches the performance of leading proprietary models, thereby making significant contributions to the field of image editing. More details please refer to our [technical report](https://arxiv.org/abs/2504.17761).
## 2. Model Usage
### 2.1 Requirements
The following table shows the requirements for running Step1X-Edit model (batch size = 1, w/o cfg distillation) to edit images:
| Model | height/width | Peak GPU Memory | 28 steps w flash-attn |
|:------------:|:------------:|:------------:|:------------:|
| Step1X-Edit | 512x512 | 42.5 GB | 5 s |
| Step1X-Edit | 768x768 | 46.5 GB | 11 s |
| Step1X-Edit | 1024x1024 | 49.8 GB | 22 s |
* The model is tested on one H800 GPUs.
* Tested operating system: Linux
* We recommend to use GPUs with 80GB of memory for better generation quality.
### 2.2 Dependencies and Installation
python >=3.10.0 and install [torch](https://pytorch.org/get-started/locally/) >= 2.2 with cuda toolkit and corresponding torchvision. We test our model using torch==2.3.1 and torch==2.5.1 with cuda-12.1.
Install requirements:
``` bash
pip install -r requirements.txt
```
Install [`flash-attn`](https://github.com/Dao-AILab/flash-attention), here we provide a script to help find the pre-built wheel suitable for your system.
```bash
python scripts/get_flash_attn.py
```
The script will generate a wheel name like `flash_attn-2.7.2.post1+cu12torch2.5cxx11abiFALSE-cp310-cp310-linux_x86_64.whl`, which could be found in [the release page of flash-attn](https://github.com/Dao-AILab/flash-attention/releases).
Then you can download the corresponding pre-built wheel and install it following the instructions in [`flash-attn`](https://github.com/Dao-AILab/flash-attention).
### 2.3 Inference Scripts
After downloading the [model weights](https://huggingface.co/stepfun-ai/Step1X-Edit), you can use the following scripts to edit images:
```
bash scripts/run_examples.sh
```
This script runs the inference code on example inputs. The results will look like:
