Krea 3
Krea 3 is Krea's instruction-following image model. It generates images from text and edits or combines up to 8 input images from a plain-language instruction: say what should change and what should stay the same. Output is 1K or 2K in ten aspect ratios plus automatic framing, with a fast and a high thinking mode.
50% less than the priciest provider (3 compared). Same model, same output.
Prices last checked November 22, 2026 at 01:00 CET: Krea 3 costs from $0.025 per image on Scenetra, billed at provider cost with no markup. Each run is priced across 3 providers and runs on the cheapest one that supports your settings. Every price is re-checked against the providers daily at 01:00 CET.
Aspect Ratio
Aspect ratio of the output. 'auto' lets the model pick the framing from your input images and prompt.
Resolution
Output resolution (1K is about 1 megapixel, 2K about 4 megapixels).
Thinking Mode
How much planning happens before generation. 'fast' suits most requests; 'high' plans complex or multi-step edits more carefully, is slower and costs more.
50% less than the priciest provider (3 compared). Same model, same output.
Your result renders here.
Use Krea 3 via the API
One REST call per generation, routed to the cheapest provider and billed in credits at provider cost. Endpoints, request fields, code examples and pricing are in the full API reference.
What makes Krea 3 different
One model for generating and editing
Krea 3 takes a prompt alone and returns an image, or takes a prompt plus your images and returns an edit. On Scenetra it is one node: leave the image input empty for text-to-image, connect images to edit. A 1K image costs $0.025 and a 2K image $0.06, billed at provider cost with 0% markup.
Edits that follow plain instructions
Write the change the way you would say it: lead with what should change, then add one sentence about what should stay the same. Krea 3 is built to keep the subject and composition of your input while applying the edit, from a small local change such as recoloring an object to a full restyle of a room.
Combine up to 8 input images
Give it a person, a product and a setting as separate images and describe how they go together. Refer to each image by what it shows, for example the sneaker from the second image on the person in the first. Each input image adds to the price (+$0.025 per input at 1K, +$0.08 at 2K).
Fast by default, High when the edit is complex
Fast thinking handles most requests in seconds. High thinking plans multi-step edits and several input images more carefully before generating; it takes longer and a 2K image then costs $0.08. Start on Fast and switch to High only when a result misses the request.
Parameters
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt* | text | Describe the image to generate, or the edit to make. For edits, say what should change and what should stay the same. | — |
| Input Images | image upload | Optional images to edit or combine (up to 8). Leave empty for text-to-image. Each input image adds to the price. | — |
| Aspect Ratio | select | Aspect ratio of the output. Auto lets the model pick the framing from your input images and prompt.auto · 1:1 · 4:5 · 3:4 · 2:3 · 9:16 · 5:4 · 4:3 · 3:2 · 16:9 · 21:9 | auto |
| Resolution | select | Output resolution (1K is about 1 megapixel, 2K about 4 megapixels).1K · 2K | 2K |
| Thinking | select | How much planning happens before generation. Fast suits most requests; High plans complex or multi-step edits more carefully, is slower and costs more.fast · high | fast |
Pricing
From $0.025 per image
Pay per generation — only pay for what you use.
1K
$0.025
per image
2K
$0.06
per image
2K, high thinking
$0.08
per image
Each input image at 1K
+$0.025
per image
Each input image at 2K
+$0.08
per image
What a video costs
| Duration | 1K | 2K |
|---|---|---|
| 1 image | $0.025 | $0.06 |
| 10 images | $0.25 | $0.60 |
| 40 images | $1.00 | $2.40 |
Use cases
E-commerce & product
- Place one packshot in styled campaign scenes
- Recolor or re-material a product
- Lifestyle scenes from a product photo
- Sketch to finished product photograph
Marketing & ads
- Campaign variations from one hero image
- Swap backgrounds and settings
- Combine a model, a product and a location
- Seasonal refreshes of existing creative
Interiors & architecture
- Reupholster furniture from a fabric swatch
- Repaint a room from colour chips
- Restyle a space while keeping the layout
- Period or mood variations of one room
Fashion & characters
- Carry one person and outfit across shots
- Outfit and accessory swaps
- Lookbook sequences from a single look
- Consistent characters for storyboards
Frequently asked questions
What is Krea 3?+
Krea 3 is Krea's instruction-following image model, released in October 2026. It generates images from text and edits or combines up to 8 input images from a plain-language instruction. On Scenetra it runs as a node in the visual workflow editor and through the API.
How much does Krea 3 cost?+
On Scenetra a 1K image costs $0.025 and a 2K image $0.06. High thinking at 2K costs $0.08. On an edit, each input image adds to the price (+$0.025 per input at 1K, +$0.08 at 2K). You pay per generation from your credit balance at provider cost with 0% markup.
Can Krea 3 edit images?+
Yes. Editing is what the model is built for. Attach one image and describe the change, or attach up to 8 and describe how to combine them. Without any input image it works as a text-to-image model.
What is the difference between Fast and High thinking?+
Before generating, Krea 3 interprets your prompt. Fast does this quickly and suits clear, simple requests. High plans complex or multi-step edits and several input images more carefully; it is slower and costs more.
What resolutions and aspect ratios does Krea 3 support?+
1K (about 1 megapixel) and 2K (about 4 megapixels), in ten aspect ratios: 1:1, 4:5, 3:4, 2:3, 9:16, 5:4, 4:3, 3:2, 16:9 and 21:9. Auto lets the model choose the framing from your input images and prompt.
How is Krea 3 different from Krea 2?+
Krea 2 Medium and Large are text-to-image models whose reference images transfer style only. Krea 3 edits the content of the images you give it and can combine several into one result. Pick Krea 2 for style-driven generation and Krea 3 when you need to change or compose existing images.
Related models
Krea 2 Large
Krea's flagship foundation image model, strongest at photorealism and rich, expressive styles. Krea 2 Large generates from text prompts with up to 10 style-reference images that transfer a look onto the output, a four-level creativity control, and eight aspect ratios up to cinematic 2.35:1.
View model →Nano Banana 2
Generate high-quality images at up to 4K resolution using Google's Gemini 3.1 Flash model. Nano Banana 2 is optimized for speed, supports text-to-image and image editing with reference images, and offers flexible aspect ratios.
View model →GPT Image 2.6 Sunburst
OpenAI's precision image model, now with animated GIF and SVG vector output. GPT Image 2.6 Sunburst generates and edits images at 1K, 2K, or 4K with five quality levels up to Max, native transparent backgrounds, up to 16 reference images for exact, instruction-following edits, and up to 4 images per run, with a planning setting for dense text and complex layouts.
View model →Start creating with Krea 3
Use Krea 3 alongside 50+ other AI models in Scenetra's visual workflow editor. No setup required.
Get Started