flux-kontext-max is a model in the Black Forest Labs Kontext series for complex image editing. It can generate new images from text or combine existing images with modification instructions to complete edits. It is suited for continuing creation around an established visual: adjusting image elements, modifying local details, and aiming for overall consistency, serving product presentations, creative design, and marketing asset iteration.
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.
The above are the creation and invocation specifications for this platform entry. Image aspect ratios do not equal fixed output pixels and do not represent the native model's resolution limit.
The focus of Kontext Max is context-aware editing: using an existing image as the starting point and describing the objects or local details that need to change through text. It is suitable for tasks where the composition is already established and only further adjustments are needed. Stating both the modification target and the elements to preserve in the instruction helps express creative intent more clearly.
When an image needs to balance the subject, background, and overall style, Kontext Max can be considered for complex editing. It aims to modify details while maintaining image consistency, making it suitable for further exploration of existing designs. In practical use, distinguish primary modifications from secondary requirements to avoid conflicting descriptions.
The same model provides both generate and edit operations. You can first use text to establish a visual draft, then use the selected image link for subsequent modifications. Completed results are delivered as image links and support asynchronous task processing, making it convenient to connect asset generation, previewing, manual selection, and continued editing.
Provide an existing product display image, describe the background, scene elements, or local visual details you want adjusted, and specify the product appearance that must be retained. The deliverables are revised display image candidates, suitable for comparing different presentation approaches; before formal use, check whether the packaging text, logos, and product structure are accurate.
Based on an already selected advertising sketch, propose specific changes, such as replacing the background atmosphere or adjusting an image element. You can organize candidate assets in landscape, square, or portrait proportions, then have designers complete the layout and finalization. It is suitable for supporting visual exploration and should not replace final brand guideline checks.
Use character, scene, or illustration concept art as input, and indicate the focus of this round of changes in text. Preserve the original and result each time, select an appropriate version, and continue editing to create a comparable creative record. The deliverables can be used for design discussions and proposal reviews, reducing the work of rebuilding images from blank prompts each time.
flux-kontext-pro and flux-kontext-max are both designed for contextual image editing. If the task involves only a clear, single change, evaluate Pro first; if you need to account for more relationships within the image or have stricter requirements for the final visual, try Max first. When selecting a model, compare results using the same original image and instructions, focusing on whether the changes are correctly made and the subject is preserved, rather than only looking at version names.
flux-kontext-max and flux-2-max are not the same model. The key reason to choose Kontext Max is to make contextual changes around existing images. If the main work is creating entirely new visuals from text, you can also compare other Flux generation models. FLUX.1 Kontext [dev] is another version, and its open weights, deployment methods, or parameters should not be directly applied to Max.
For text-to-image, specify the subject, materials, lighting, and canvas ratio; for modifications to existing images, prepare image_url and clearly state what to change and preserve.
Submit model=flux-kontext-max, action=generate, prompt, and size=16:9 to /flux/images; when editing images, select edit and add the image URL, while count is used only 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.
Keep the architecture, character positions, and clothing from the reference image, changing only the sky to blue hour and adjusting the window lighting to create a photorealistic advertising photography look.
Check whether the subject identity, window structure, and perspective are preserved; highly demanding details should be reviewed by area rather than comparing only the overall atmosphere.
No. It supports generate for text-to-image generation and edit for image editing. For generation, describe the desired image; for editing, provide image_url and modification instructions. Both operations should submit action, prompt, and size, and explicitly set model to flux-kontext-max.
First describe the object to modify, then the desired effect, and add the subject, composition, or style you want to preserve. For example, change the background to a studio setting while keeping the product appearance. Complex tasks can be split into different candidates for comparison to avoid putting conflicting requirements into a single instruction.
Pass the image ratio through size, such as 1:1 for square images, 16:9 for landscape images, and 9:16 for portrait images. This controls the ratio rather than the exact pixel dimensions. Even for editing tasks, it is recommended to explicitly submit the ratio to help maintain the required output aspect ratio.
Both are suitable for editing based on an original image and text. For routine modifications, compare Pro first; for complex editing, prioritize evaluating Max. It is best to test with the same input and observe modification accuracy and overall consistency; the Max name does not mean every original image and every instruction will necessarily produce more suitable results.
count does not apply to editing tasks; multiple editing candidates should be submitted separately. For background processing, use async=true, query the result after obtaining task_id, or receive completion notifications through callback_url. Finally, read image_url from the returned data for previewing or downloading.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗
↗Pricing
$0.0751· per generation
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs