flux-2-flex is an image generation and editing model in the Black Forest Labs FLUX.2 series that emphasizes typography and detail preservation. It is suited for composing titles, labels, and subjects into a single image, rather than merely generating backgrounds. With text descriptions or existing images combined with editing instructions, it can create posters, packaging concepts, UI visual mockups, and product scene images.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudflux-2-flexInput 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.
FLUX.2 [flex] is natively focused on typography and detail, while this platform provides workflows for generation by aspect ratio and editing via image links.
flex stands out for typography, making it suitable for designing titles, short labels, and visual subjects together. Prompts can clearly specify the text to display, its placement, and its hierarchy, allowing the model to organize the image around the content. Compared with creations that only pursue atmosphere, it is better suited for visual tasks where text itself carries information.
When packaging outlines, decorative patterns, or label details determine whether a design works, flex's detail-preservation orientation becomes more valuable. You can describe the elements that need to be retained, then provide modification goals for the background, materials, or style; this approach is suitable for continuing to explore an existing design rather than regenerating from scratch every time.
The same model can generate images from text concepts or edit existing images using text instructions. Establishing the composition and aspect ratio first, then refining candidate images, can organize exploration into a continuous process. Results are delivered as image links and support asynchronous processing, making them easy to integrate into asset review and backend task workflows.
Enter the event theme, exact title, brand colors, and subject description to create poster candidates containing text. Landscape and portrait content can each use the corresponding aspect ratio, with title placement and whitespace specified in the prompt. The deliverables are suitable as a design starting point, followed by text proofreading, layout refinement, and checks before official release.
Provide product images, explain which labels, shapes, and key details need to be preserved, then propose a new background or presentation style. This can be used to explore packaging visuals, product displays, and promotional scenes, keeping design discussions focused on visible results. For real products, pay close attention to whether branding, structure, and patterns have changed.
Describe interface sections, headings, buttons, and information hierarchy in text to generate visual concepts for discussion. flex's layout orientation is suitable for observing how text and graphics organize information together, but the deliverables are still images, not functional interfaces; after confirming the direction, components, interactions, and layout specifications still need to be completed in design tools.
If the acceptance criteria focus on whether titles can integrate into the composition, labels are clear, and small decorative elements can be preserved, consider flex first. FLUX.2 [pro] is more geared toward general production-grade generation and editing, making it suitable for everyday visual tasks across a wide range of subjects. The choice between them should be based on actual sample images: whether specialized typography and detail performance are needed, or overall generation and editing capabilities matter more.
When a task places greater emphasis on consistency in complex revisions, strict instruction following, and multiple style expressions, consider FLUX.2 [max]; if the focus is on quickly exploring a large number of drafts, FLUX.2 [klein] is better positioned for prototype iteration. The reason to choose flex is text and detail, rather than treating it as the universal answer with the highest image quality or fastest speed for all tasks.
For text-to-image, specify the subject, materials, lighting, and aspect ratio; for modifying an existing image, prepare image_url and clearly state what to change and preserve.
Submit model=flux-2-flex, action=generate, prompt, and size=16:9 to /flux/images; when editing an image, choose edit and add the image URL. count is only used for generation.
Save the returned image_url; for asynchronous operations, query using task_id or configure callback_url. Compare key elements before and after modification, then use the completed image in subsequent design work.
Create a clean book club visual, with the main title “Read Together”, the subtitle “Stories connect us”, a stack of books and a warm-toned desk lamp as the main subject, layered text sizes, and a light background.
Check spelling, letter spacing, and hierarchy in the order of the title and subtitle, and zoom in to inspect small text; typefaces and layouts requiring precision still need to be completed manually.
flex is better suited for tasks where typography and small details are key acceptance criteria, such as poster headlines, packaging labels, and UI concept designs. pro is more geared toward general production-grade generation and editing. It is recommended to compare results using the same set of representative tasks rather than deciding which is more suitable based solely on the model name.
Yes. Use model set to flux-2-flex and action set to edit, and provide image_url, an edit prompt, and size in aspect-ratio format. In your description, clearly specify both the areas to modify and the content to preserve. After obtaining the result, check whether the text, outlines, and details meet your goals.
The size for this model should be specified as an aspect ratio, for example 1:1 for square images, 16:9 for landscape images, and 9:16 for portrait images, rather than a pixel-dimension string. The ratio determines the image shape; if fixed pixel dimensions are needed later, you can resize the result after obtaining it.
Clearly list the text that needs to appear, and specify the placement and visual hierarchy of the main headline and supporting labels. Avoid cramming large amounts of text into overly small areas, and test the key headline first. After generation, still proofread the typeface, spelling, and numbers; important text can be retypeset during the final design stage.
For generation tasks, you can use count to specify the quantity, submit asynchronous tasks through async, or set callback_url to receive completed results. Once you have the image links, proceed with downloading, review, and archiving. Editing tasks do not use count; submit each image to be processed and its edit instructions separately.
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
