Dual-Mode AI Editor:
Combines visual-semantic control via Qwen2.5-VL and appearance control through VAE Encoder for versatile edits.
Qwen-Image-Edit is an open-source image editing model based on Qwen-Image, supporting semantic and appearance editing with precise, visually coherent results. It also handles bilingual (Chinese and English) text editing while preserving font, size, and style, making it a versatile tool for advanced visual content manipulation.
The prompt to generate the image with
Click to upload or drag and drop
Supported formats: JPEG, PNG, WEBP Maximum file size: 10MB
The URL of the image to edit.
Acceleration level for image generation. Options: 'none', 'regular'. Higher acceleration increases speed. 'regular' balances speed and quality. Default value: "none"
The size of the generated image. Default value: landscape_4_3
The number of inference steps to perform. Default value: 30
The same seed and the same prompt given to the same version of the model will output the same image every time.
The CFG (Classifier Free Guidance) scale is a measure of how close you want the model to stick to your prompt when looking for a related image to show you. Default value: 4
If set to true, the function will wait for the image to be generated and uploaded before returning the response. This will increase the latency of the function but it allows you to get the image directly in the response without going through the CDN.
If set to true, the safety checker will be enabled. Default value: true
The format of the generated image. Default value: "png"
The negative prompt for the generation Default value: " "

Explore different use cases and parameter configurations
Complete guide to using
💡 Note: Price ≈ $0.0165 per megapixel. Our credits are valued at $0.005 per credit. The table above shows the cost for one image — for multiple images, simply multiply the credits or cost by the number of images.
| Category | Resolution (px) | Total Pixels | Credits Used | Actual Cost (USD) |
|---|---|---|---|---|
| Square 1:1 | 512 × 512 | 262,144 | 1 | 0.005 |
| Square HD | 1024 × 1024 | 1,048,576 | 3.5 | 0.0175 |
| Portrait 3:4 | 768 × 1024 | 786,432 | 2.5 | 0.0125 |
| Portrait 9:16 | 576 × 1024 | 589,824 | 2 | 0.01 |
| Landscape 4:3 | 1024 × 768 | 786,432 | 2.5 | 0.0125 |
| Landscape 16:9 | 1024 × 576 | 589,824 | 2 | 0.01 |
Discover Qwen Image Edit API, the open-source AI tool from Alibaba's Qwen team. Unlock precise text rendering, semantic transformations, and appearance controls for seamless edits. Elevate your creative workflow today.

Qwen Image Edit API is an advanced open-source image editing foundation model developed by Alibaba's Qwen team, extending the capabilities of the 20B Qwen-Image model.
Combines visual-semantic control via Qwen2.5-VL and appearance control through VAE Encoder for versatile edits.
Supports precise editing of English and Chinese text, matching original fonts, sizes, and styles seamlessly.
Fully accessible under Apache 2.0 license, enabling free use, fine-tuning, and deployment on platforms like Hugging Face.
New dual-path processing for high-level changes like style transfers and pixel-accurate tweaks like object removal.
Superior handling of complex calligraphy and multi-language text, with flawless integration into images.
Sets new SOTA results on public editing datasets, surpassing open models in style consistency and efficiency.
Qwen Image Edit's core strength lies in its dual editing modes, blending semantic understanding with appearance precision for magical results.
Effortlessly add, edit, or delete bilingual text while preserving original aesthetics – a game-changer for posters and graphics.
Handles English, Chinese, and more with native rendering, ensuring cultural accuracy in global designs.
Generate or edit high-quality images in seconds, optimized for real-time applications on various platforms.
Fine-tune aspect ratios, poses, and layouts with intuitive prompts for professional-level outputs.
Add or remove elements like backgrounds or objects in product photos, ideal for e-commerce enhancements.
Get started with our product in just a few simple steps...
Upload an image and craft a prompt describing edits, e.g., "Add a red hat to the character" or "Change background to beach."
Customize settings: select acceleration , image size , inference steps, seed for consistency, CFG scale for prompt adherence, safety checker , output format (PNG), and wait option for immediate results.
Click generate. View and download the edited image.