FLUX.1 Kontext Max is a flat $0.0715 per image on HiAPI, or 143 credits, and image editing costs exactly the same as text-to-image. A $10 top-up buys about 139 images, whatever the aspect ratio, output format or number of edit rounds.
FLUX.1 Kontext Max is live on HiAPI. Open the model page to generate right away; new accounts get 200 credits on sign-up.
What does one FLUX.1 Kontext Max image cost?
| Item | Price |
|---|---|
| Single image | $0.0715 (143 credits) |
| Text-to-image | $0.0715 |
Image edit (with input_image) | $0.0715 |
| $10 budget | about 139 images |
| 200 new-user credits | $0.10, about 1.4 images |
You pay per successfully delivered image, one image per request, and the amount held for a failed task is refunded automatically. The model page shows prices in credits, with 2,000 credits = $1.
A first test costs $10 for about 139 images. That is enough to take a prompt from rough draft to final and still run a small series. Put a $10 budget to work on the model page →
What do 100, 1,000 and 10,000 images cost?
| Images per month | Budget |
|---|---|
| 100 | $7.15 |
| 1,000 | $71.50 |
| 10,000 | $715 |
Your bill is the image count times $0.0715. All six aspect ratios (21:9, 16:9, 4:3, 1:1, 3:4, 9:16) cost the same, and so do jpeg and png, so the budget stays a single line. A thousand edited images a month runs $71.50; ten thousand runs $715.
How do you get more out of each image?
Test small, then batch. Confirm composition, palette and text placement with three to five images before committing to a full set. The more specific the prompt, the fewer retries you pay for — and at a flat rate, every discarded image is a full $0.0715.
Edits do not cost extra. Pass input_image and you are in edit mode at the same price as generation. Leave aspect_ratio empty and the output follows the source image, with pixels nudged to a multiple of 32 (1280×720 becomes 1328×752), so you avoid a wasted run just to match dimensions. Anything you do not mention is preserved, and one request can change material, add an element and adjust lighting together.
Submit each task once. Poll with the returned taskId, or receive the final state through callback.url. A repeated POST creates a new task and is billed again.
Send one source image. input_image takes a single image, which suits edits that start from one original. When a task needs several references at once, use GPT Image 2 image editing instead.
The edit below swaps the linen armchair for deep forest-green velvet and adds a slim brass floor lamp, while keeping the window, books, cup, table and afternoon light untouched. It costs the same $0.0715:
How does the price compare with GPT Image 2 image editing?
| Model | Per image | Reference images | Aspect ratios |
|---|---|---|---|
| FLUX.1 Kontext Max | $0.0715 flat | 1 | 6 |
| GPT Image 2 image editing | $0.03 at 1K, $0.04 at 2K, $0.06 at 4K | 1–16 | 16 |
| Qwen Image 3.0 Pro image editing | $0.0497 at 1K, $0.0993 at 2K | multiple | 1K and 2K |
At 1K, one FLUX.1 Kontext Max image costs about as much as 2.4 GPT Image 2 edits; at 4K, about 1.2. The choice follows the job rather than the sticker price: reach for FLUX.1 Kontext Max when you are editing a single image and need the original proportions and overall consistency held together, with material, light and on-image text changing in one pass. Reach for GPT Image 2 when one request has to carry several references, or when you need a transparent background at 1K.
FAQ
Does an edit cost the same as a text-to-image generation?
Yes. input_image switches the model into edit mode and the price stays $0.0715 per image, with no premium for the number of changes.
Am I charged for failed generations? No. The amount held for a failed task is refunded in full, automatically.
Do different aspect ratios or PNG output cost more? No. All six ratios and both output formats share one price; only the image count changes the bill.
Can I try it for free? Yes. New accounts get 200 credits, about $0.10, which covers one image with a little left over. Generate it straight from the model page.
Generate your first image now
A Lisbon travel poster prompt is ready to go: open the model page, set the aspect ratio to 3:4 and hit Run. The headline type, tram and tiled facades all land in one pass, for $0.0715.
