Use FLUX.1 Kontext online by selecting Kontext pro or max in BFL Playground, uploading an image, and describing the edit. Black Forest Labs also provides these hosted models through its API, with the same published per-image pricing as Playground. Playground guide, Pricing
This walkthrough covers the online process from choosing a paid variant to saving the result and transferring settings into code. For the model family's broader capabilities, see the Kontext overview.
What are the key facts about FLUX.1 Kontext online?
| Field | Verified detail |
|---|---|
| Developer | Black Forest Labs. Announcement |
| Released | Kontext pro/max and BFL Playground: May 29, 2025. Announcement |
| Type | Hosted image generation and instruction-based editing. Overview |
| Size or parameters | Hosted pro/max: not published in the cited overview; separate dev checkpoint: 12B. Overview, Dev card |
| License and access | Pro/max: hosted services with no open weights; dev: downloadable weights under a non-commercial model license. Official repository, Dev card |
| Where it runs | Pro/max run on provider infrastructure, accessed through Playground or REST API. Overview |
| Price | Pro: $0.04 per image; max: $0.08 per image, for API and Playground. Pricing |
How much does FLUX.1 Kontext cost online?
BFL lists Kontext pro at four credits per image and Kontext max at eight credits, with one credit equal to $0.01 USD. These are usage charges for the selected variant. Pricing checked September 5, 2026. Pricing
| Hosted variant | Price per image | Illustrative cost for 25 images |
|---|---|---|
| Kontext pro | $0.04 | $1.00 |
| Kontext max | $0.08 | $2.00 |
The totals are arithmetic using BFL's published rates. They describe 25 generated images, regardless of how many you choose to keep. Pricing
Set an evaluation budget around candidate images rather than finished assets. If a project needs several attempts for each accepted edit, record those attempts when estimating the next project. Keep promotional credits separate from the recurring rate you are comparing.
Use the same input image and editing task when evaluating pro against max. Decide whether the requested change worked and whether the subject, background, and lettering still meet your brief before choosing a variant for more requests.
What upload and output limits affect Kontext online editing?
The Kontext editing API accepts a base64-encoded image or an image URL. Its documented input limit is 20MB or 20 megapixels; outputs are approximately one megapixel. A larger input does not therefore establish an equal-resolution output. Editing parameters
The API supports output aspect ratios from 3:7 to 7:3. Check the aspect ratio and intended crop before spending requests on a layout that needs a different shape. Editing parameters
BFL recommends FLUX.2 for new generation and editing projects and retains Kontext as a previous-generation option. Check that Kontext is the selected model when following this workflow. Overview
Kontext's authors document editing failures and identity degradation. Save the original input and review every accepted revision against it, especially where a product outline, face, or printed label must remain recognizable. Technical report
How do you edit and download images in BFL Playground?
Select Kontext and upload the starting image
Open BFL Playground and sign in. Select FLUX.1 Kontext pro or max, upload your source image, and enter the edit instruction. BFL's help guide documents these model choices and the image-upload workflow. Playground guide, Account setup
Start with one clearly visible subject and a simple change. Keep a copy of the original image, and name the output files by the revision they contain so you can retrace the sequence later.
For example, try: “Change the chair's upholstery to dark green fabric. Keep the chair shape, wooden legs, camera angle, and surrounding room unchanged.” This is an original practice instruction for evaluating the workflow.
Generate, review, and save the result
Click Generate and inspect the output. Use the download icon on the generated image to save it to your device. Playground guide
For the chair example, compare the upholstery color first, then check the frame, legs, shadows, and nearby furniture. A successful recoloring still needs review if another required detail has changed.
Keep an accepted version as a separate file before starting another edit. If a later result changes the chair's outline, return to the last usable image and simplify the next instruction. Record rejected results as well as accepted ones when reviewing the experiment's cost.
Export the successful generation's settings
BFL documents an API Code control at the top right of a successful generation. Choose Python, JavaScript, or cURL to copy a snippet containing that generation's parameters. Playground guide
Use the exported code as a reference when moving the task into an application. Check the selected model, input image, prompt, and output options together so the application submits the configuration you intended to evaluate.
How do you submit a Kontext online edit through the API?
Create a BFL account, add credits, and create a project API key using the quickstart. Store the key in your environment as BFL_API_KEY and keep it out of client-side code. Quickstart
Install Python's requests package and save your input as chair.png. This example follows BFL's base64 input pattern, submits a Kontext pro edit, and prints its polling URL. API guide
import base64
import os
from pathlib import Path
import requests
response = requests.post(
"https://api.bfl.ai/v1/flux-kontext-pro",
headers={"x-key": os.environ["BFL_API_KEY"]},
json={
"prompt": "Change the chair upholstery to dark green fabric.",
"input_image": base64.b64encode(Path("chair.png").read_bytes()).decode(),
},
timeout=60,
)
response.raise_for_status()
print(response.json()["polling_url"])
Send GET requests to the returned polling URL with the same API key. When the status is Ready, retrieve the image at result.sample; stop and inspect the response if the task reports Error or Failed. API guide
BFL's signed result URLs remain valid for ten minutes. Download the image promptly and preserve its task identifier, input, prompt, and selected model together. Result retrieval
How do Playground, API, and local Kontext access compare?
| Route | Workflow distinction |
|---|---|
| Playground | Select hosted pro/max, upload, edit, download, and export API code through the browser. Help guide |
| BFL API | Submit hosted requests from code and retrieve completed images through polling. API guide |
| Kontext dev | Download a separate editing checkpoint and provide a compatible local runtime. Dev card |
Use the browser to review an individual task, or the API when your application needs to submit and track requests. For local execution, follow the Kontext open-weights guide and review the dev checkpoint's own setup and license.
What should you know before using Kontext online?
Can I use FLUX.1 Kontext without installing model weights?
BFL Playground offers hosted FLUX.1 Kontext pro and max in the browser. Select either variant and upload an image for editing. Playground guide
Does Kontext cost more in Playground than through the API?
BFL lists the same Kontext prices for Playground and API: $0.04 per pro image and $0.08 per max image. Select the same variant when comparing those access paths. Pricing
How do I save a Kontext image from Playground?
Click the download icon on the generated Kontext image to save it to your device. BFL says image downloads are available immediately after generation. Playground guide
Can I copy a Playground edit into API code?
BFL Playground provides API Code for successful generations, with Python, JavaScript, and cURL options. The exported snippet includes the generation's parameters for use in an application. Playground guide
Sources
- Kontext and Playground announcement
- Kontext variants and access
- Kontext image-editing API guide
- Kontext dev model card
- BFL pricing
- BFL account and API setup
- Kontext technical report
- BFL Playground controls and code export
- BFL hosted and open-weight model access
Top comments (0)