qwen-image-3.0-pro is Alibaba's Qwen Image 3.0 Pro image model, designed to turn detailed creative requirements into visually rich, clearly structured finished images. It balances multi-region layouts, multilingual text, and subtle material rendering. On this platform, you can use a single entry point for text-to-image generation, reference image editing, and batch delivery, making it especially suitable for posters, menus, storyboards, and product concept design.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
Native capabilities describe the model's creative scope; use the platform entry-point instructions for reference image count, dimension settings, and task delivery methods. Token and character limits are calculated separately.
The focus of Pro is not just generating backgrounds, but organizing titles, body text, illustrations, and multiple content areas within the same image. You can specify layout sections, text hierarchy, and exact copy separately in the prompt to create menus, newspaper-style posters, or storyboard pages, reducing the work of arranging all the text again after creating the image.
The model emphasizes details such as micro-expressions, pores, and strands of hair, making it suitable for precise visual requirements for people and products. When creating, describe the subject, materials, lighting, and composition separately instead of relying only on abstract terms like “premium feel,” so the direction of the final image is clearer; important details should still be checked at original size.
After adding reference images, you can modify the background, colors, style, or local elements around an existing subject. For multi-image tasks, clearly specify whether each image is responsible for the subject, composition, or style, and list the features that must be retained. Generation and editing share the same interface, with results delivered through image URLs for continued display, download, and distribution.
Enter approved dish names, campaign copy, brand colors, and regional layouts to generate menus or promotional posters with images and text. When product appearance needs to be retained, add reference images and then explore different compositions in batches. Before delivery, check prices, spelling, and text alignment item by item, and keep suitable versions for further refinement.
Organize key knowledge points, question text, or shot sequences into section-based prompts to generate exam-style pages, diagrams, or multi-panel storyboards. Clearly defining the subject, explanation, and sequence of each panel makes complex content easier to review. Generated images can be used as discussion drafts, but formulas, labels, and shot continuity should be verified separately.
Provide product reference images, target audiences, and interface content to create visual concepts for websites, games, or livestream interfaces. Use specific sections to describe navigation, display areas, and button copy, creating full-page images for discussion; it delivers visual proposals, not functional websites or applications with real interactions.
When tasks involve complex layouts, dense text, or business details that require careful viewing, Pro is better suited for high-precision final output. If the main goal is frequently trying styles, exploring compositions, and iterating in batches, consider qwen-image-3.0 from the same series. Both can use similar workflows, but you should explicitly specify the corresponding model ID to avoid treating Standard as Pro.
If the core task is to generate complete copy and visuals together, first use a structured prompt to determine the layout, then refine the details; if an existing image needs to preserve the subject, add a reference image and clearly specify what must remain unchanged. For projects requiring editable text, precise vector graphics, or functional interfaces, you can use the Pro result as a visual draft, then complete the final production with design or development tools.
Organize the title, body text, and layout; when editing is needed, prepare 1–3 publicly accessible images and separately describe the subject, background, and style references.
Send model=qwen-image-3.0-pro and prompt to /qwen-image/images; start with size=2048*2048 and n=1. When editing, use image_urls; do not write dimensions separated by x.
Read image results synchronously, or use async=true to save task_id and then query /qwen-image/tasks; verify the text and subject, then use the selected image for the next editing round.
Create a one-page coffee menu, with “Espresso 浓缩” and “Latte 拿铁” arranged by category on the left and corresponding prices on the right; use only the provided copy, with off-white and dark brown as the main colors, plus a small cup illustration.
Check the Chinese and English, numbers, and row-column alignment item by item, then check whether the cup and text are crowded; dense small text still needs to be reviewed in design software.
When submitting a request to POST /qwen-image/images, explicitly set model to qwen-image-3.0-pro and provide a prompt. Do not omit the model name or use qwen-image-3.0 instead; the latter is a Standard model in the same series, suited to different creative trade-offs.
Text-to-image generation is performed when image_urls is not provided; for editing, you can provide 1–3 publicly accessible images. Specify the purpose of each image in the prompt, for example, preserve the subject from the first image and reference the composition of the second, and state which features must not change to avoid attributes from different images being mixed together.
Use a width*height pixel format for size, for example 1024*1024; the aspect ratio must be between 1:8 and 8:1. n can be set to 1–6, and one image is generated by default. Text-dense designs should be checked for readability at the delivery size rather than simply increasing the batch quantity.
First describe the work type, then list the exact copy, font direction, and visual requirements by area. Native long-instruction capability and platform character limits use different units and cannot be directly converted. Multilingual content should provide approved text; language support also does not mean that all font and small-text combinations can be rendered reliably.
You can receive results synchronously, or set async to true to obtain a task_id, then query through /qwen-image/tasks or receive a completion callback. Successful results include image_url and are stored on this platform's CDN, making them convenient for business systems to display, download, and distribute later.
Choose a development approach or application suited to this model and follow its setup guide.
Python SDK Integration Guide
SDK installation, authentication and examplesTypeScript SDK Integration Guide
SDK installation, authentication and examples
