Image model · Image editing
FLUX Kontext Dev on Klingobanana
FLUX Kontext Dev edits an existing picture from a text instruction: change an object, swap a style, move the scene. It needs a source image and does not generate from text alone.
From 1 credits per image
Specs
- Type
- Image · Image editing
- Input
- Requires an image
- Reference images
- Up to 1
- Speed
- Medium
Price per generation
| Option | Cost |
|---|---|
| One image | 1 credits |
Credits come from your plan. See all models to compare image prices.
What FLUX Kontext Dev is good at
FLUX Kontext Dev is an instruction-following image editor from Black Forest Labs. You upload a picture, write what should change, and it returns the same picture with that change applied. That makes it a different kind of tool from a text-to-image model: the image you start with is the point, and the prompt is a patch, not a description.
It is also one of the cheapest edit options on the platform. For small, surgical jobs such as swapping a garment colour, changing the weather, or restyling a photo into an illustration, it avoids the cost and the randomness of regenerating the whole scene. If the subject, framing and light already work, you keep them.
Ideal use cases
- Wardrobe and colour swaps. Keep the pose and face, change the jacket, the sofa or the car colour. Useful for catalogue variants and for testing which look suits a character.
- Restyling a photo. Turn a straight photo into watercolour, pencil, or anime-style art while the composition stays recognisable.
- Fixing an almost-good image. An output you like except for one detail is a perfect edit job: remove the stray object, change the sky, brighten the room.
Limits to know about
- It needs a source image; without one the model has nothing to work from.
- It takes a single reference image, so it cannot merge a person from one photo with a product from another. Nano Banana 2 Edit handles multiple references.
- Very large changes (a totally different scene) drift from the original. Edits work best when most of the image stays put.
Example prompts for FLUX Kontext Dev
Weather change
Change the weather to heavy snowfall, keep the buildings, the parked cars and the pedestrians exactly as they are.
Why it works. The prompt names one change and explicitly lists what must stay, which is how you protect the rest of the frame.
Material swap
Replace the wooden dining table with a white marble table. Keep the chairs, the vase and the lighting unchanged.
Why it works. Material swaps are a strong use of editors: a clear before and after noun, with the surroundings anchored.
Style conversion
Redraw this photo as a soft watercolour illustration with visible paper texture, keep the same composition and the same facial features.
Why it works. Style transfers need the instruction to say what carries over (composition, face), otherwise the subject can change.
Settings and what costs more
- One image costs 1 credits, the lowest price among the image models.
- Prompt length and aspect ratio do not change the price, so spend your effort on the prompt rather than on shaving words.
- Input quality is the biggest lever. A sharp, well-lit source gives a clean edit; a blurry one gives a blurry edit.
- Keep the output aspect ratio the same as your source to avoid unwanted cropping of the subject.
- Make one edit per pass. Chain several small edits instead of one long instruction with four changes in it.
Compare FLUX Kontext Dev with
- Nano Banana 2 Edit (3 credits per image)Choose it for the better choice when you need several reference images or a consistent face.
- Gemini 3 Pro Edit (5 credits per image)Choose it for for edits where fine detail and realism matter more than cost.
Our take
Pick Kontext Dev for fast, low-cost surgical edits. Move to a Gemini edit model when the job needs multiple references or higher fidelity.
Use it for
- Building a consistent AI influencer, using reference images to keep one face across content.
- UGC-style product ads, from script to finished clip.
- Choosing a platform? Klingobanana vs Higgsfield, Klingobanana vs OpenArt.
Related models
Frequently asked questions
What is the difference between FLUX Kontext Dev and a normal text-to-image model?
- A text-to-image model invents a picture from words. Kontext Dev starts from your picture and changes only what you ask for, so the original subject, framing and lighting survive.
How should I write an edit prompt for FLUX Kontext Dev?
- State the change directly, then list what must not change. "Make the sweater green, keep the face and background identical" works better than re-describing the whole scene.
Can it change text inside an image?
- Short, simple text can sometimes be changed, but it is not reliable. For anything legible you must hit exactly, edit the text in a design tool afterwards.
Does it work through the API?
- Yes. Send the image URL and the instruction through the Klingobanana API with this model id and the edited image URL is returned when the job finishes.
How many credits does FLUX Kontext Dev cost?
- One image costs 1 credits on Klingobanana. The price table on this page lists every supported length.
Can I use reference images with FLUX Kontext Dev?
- Yes, it accepts up to 1 reference image.
Do I need a subscription?
- Yes. There is no free tier; credits come from a subscription plan or a top-up.
Generate with FLUX Kontext Dev
Subscribe, open the studio and pick FLUX Kontext Dev from the model list. Credits are only spent when you generate.
Generate with FLUX Kontext DevA subscription is required before generating. Plans from €15/month, cancel anytime.