How to Use Flux Kontext for Image Editing
How to access and use Flux Kontext, Black Forest Labs' image editing model, from the BFL Playground and partner platforms to the 8frame canvas. Steps and cost.
Flux Kontext is the model to reach for when you have an image that is almost right and you need to change one thing without regenerating the whole frame. This guide covers how to access it, from Black Forest Labs' own Playground and its many partner platforms to running it directly on the 8frame canvas, plus a step-by-step editing workflow and what it costs.
What you'll end up with is precise, iterative edits: swap a background, change a product color, remove an object, restyle a scene, while keeping everything else, including a character's face, exactly as it was. That in-place consistency is the whole point, and it's what separates Kontext from image models that redraw the frame every time you ask for a tweak.
What Flux Kontext is
Flux Kontext is a text-and-image editing model from Black Forest Labs. You give it an existing image plus an instruction, and it edits the image while preserving character consistency, layout, and the parts you didn't ask to change. It's built for iteration: you can stack edit on edit, refining step by step with low latency and without quality drift across passes. For the wider Flux family context, see What is Flux. Black Forest Labs released FLUX.2 in November 2025 and now points new projects there, while continuing to support Kontext in production as the previous generation.
Official access routes
Flux Kontext is a Black Forest Labs (BFL) model, and BFL distributes it broadly rather than locking it to one app.
The BFL Playground. You can test Kontext directly at playground.bfl.ai, the first-party way to try the model without a partner account.
The BFL API. For developers, Kontext is available through the BFL API, documented for programmatic editing at scale.
Partner platforms. This is where most people actually meet Kontext. It's integrated into creative tools including Krea, Freepik, Leonardo, OpenArt, and Lightricks, and offered by infrastructure partners such as FAL, Replicate, Runware, DataCrunch, TogetherAI, and Comfy. Each wraps the model in its own UI and pricing.
Open weights. BFL released a FLUX.1 Kontext [dev] open-weights variant on Hugging Face, so you can run it locally if you have a capable GPU and want full control. The open variant trades the convenience and speed of the hosted endpoints for self-hosting overhead.
The spread of routes is a genuine strength: Kontext is almost everywhere. The friction is that each platform prices and rate-limits it differently, and none of them put it on the same surface as the video model you'll animate the edited frame with.
How to use Flux Kontext on the 8frame canvas
On 8frame, Flux Kontext is an editing node that sits next to the generation and video models, so an edit and the clip it feeds live in one place.
-
Drop your source image on the canvas. This is the frame you want to change: a product photo, a generated still, a character shot. Wire it into a Flux Kontext node.
-
Write a single, specific edit instruction. Kontext rewards one clear change at a time. Good: "replace the plain background with a sunlit marble kitchen counter, keep the product, label, and lighting on the bottle unchanged." Vague instructions produce vague edits. For a full set of tested edit prompts, see Flux Kontext prompts for editing existing images.
-
Stack edits iteratively. This is where Kontext shines. Chain a second node onto the output to make the next change ("now add soft morning light from the left"), then a third. Because it preserves what you didn't touch, faces and product geometry hold across the whole chain instead of drifting.
-
Send the finished frame into video. Once the still is right, wire it into a Seedance or Higgsfield node to animate it, or a Topaz node to upscale. The edited frame never leaves the canvas.
Flux Kontext runs at the low end of image pricing on 8frame, in line with the Flux family (Flux Klein sits at about $0.01 per image), so heavy iteration stays cheap.
Kontext vs Nano Banana Pro for editing
Both models edit in place, so which do you use? Nano Banana Pro reasons hard about layout and text, which makes it the pick when the edit involves legible copy, badges, or packaging labels. Flux Kontext is built specifically for iterative, low-latency edit chains with strong character and structure preservation, so it wins when you're making a series of surgical changes to the same frame. On 8frame you don't have to choose in advance; drop both nodes and keep the better result. See how to use Nano Banana Pro for that side.
Official vs canvas: which to pick
| Your situation | Best route |
|---|---|
| Quick one-off test of the model | BFL Playground |
| Building editing into your own product | BFL API or an infra partner |
| You want full local control, have a GPU | FLUX.1 Kontext [dev] open weights |
| You already work inside Krea / Freepik / Leonardo | That platform's integration |
| Edits that feed into a video pipeline | 8frame canvas |
| Cheap, heavy iterative editing on a canvas | 8frame canvas |
Cost math
Say you're localizing one hero product image into 5 market variants, each needing a background swap and a headline-language change.
- On 8frame: 5 background edits through Flux Kontext at roughly a cent or two each, plus 5 text edits through Nano Banana Pro at $0.04 to $0.08, lands under $0.50 total, all on one canvas.
- A designer doing this by hand in a photo editor is 30 to 60 minutes of work per variant, or two to three hours for the set.
The editing itself is close to free wherever you run Kontext. The canvas advantage is the edit chain plus the fact that the finished frame is one wire away from becoming a video.
FAQ
What is Flux Kontext used for?
Flux Kontext is used for editing existing images: swapping backgrounds, changing colors, removing or adding objects, and restyling scenes while preserving the parts you didn't ask to change. Its defining strength is character and structure consistency across iterative edits, so faces and product geometry hold from one edit to the next.
How is Flux Kontext different from regular Flux?
Standard Flux models are primarily text-to-image generators. Flux Kontext is an editing model that takes an existing image plus an instruction and modifies it in place, keeping unspecified elements intact. Kontext is the one you reach for when you want to change part of an image rather than create a new one.
Can I run Flux Kontext for free?
You can test it at the BFL Playground, and Black Forest Labs released an open-weights FLUX.1 Kontext [dev] variant on Hugging Face that you can run locally at no per-image cost if you have a capable GPU. Hosted routes, including 8frame, charge a small per-image fee for the convenience of no setup.
Is Flux Kontext still current in 2026?
Black Forest Labs released FLUX.2 in November 2025 and points new projects there, but it continues to support Kontext in production. For iterative in-place editing with strong consistency, Kontext remains a very capable and widely available choice.
Have an image that's almost right? Edit it with Flux Kontext on the 8frame canvas, one wire away from the video model that animates it. See the full model lineup for where Kontext fits your workflow.