---
title: "Krea 3"
description: "Edit or combine up to eight reference images into one 1K or 2K image."
seo:
  title: "Krea 3 API Documentation: Parameters & Examples"
  description: "Krea 3 API documentation: model ID krea/krea-3, input modes, parameters, defaults and request examples. Edit or combine up to eight reference images into one 1K or 2K image."
---

{/* Generated by scripts/generate-model-pages.mjs from openapi/model-contracts.json. Do not edit; write notes in docs/models/_notes/krea/krea-3.mdx */}

| Model ID | Alias | Endpoint |
| --- | --- | --- |
| `krea/krea-3` | `krea-3` | [Image generation](/docs/guides/image-generation) |

## Input modes

| Mode | References | Reference roles |
| --- | --- | --- |
| `image` Image input | 1 to 8 | `reference_image` |

## Parameters

| Parameter | Description | Values | Default | Required |
| --- | --- | --- | --- | --- |
| `resolution` | Resolution: Output resolution. | `1K`, `2K` | `1K` | No |
| `aspect_ratio` | Aspect ratio: Output aspect ratio. | `1:1`, `4:3`, `3:4`, `16:9`, `9:16` | `1:1` | No |
| `n` | Outputs per request: Number of outputs from one API request (one Job), not the application batch size. For multiple Jobs, submit separate requests; n=1 does not prevent application-managed batching. | `1` | `1` | No |

## Examples

`image` input:

```json
{
  "model": "krea/krea-3",
  "input": {
    "type": "image",
    "prompt": "A watercolor fox reading a map in a pine forest",
    "references": [
      {
        "role": "reference_image",
        "file_id": "file_..."
      }
    ]
  },
  "parameters": {
    "resolution": "1K",
    "aspect_ratio": "1:1",
    "n": 1
  }
}
```

The authoritative contract is `GET /v1/models/krea/krea-3?view=contract`. A valid parameter does not mean every channel can run it, so [quote](/docs/concepts/billing#get-a-quote) the complete request. Prices and channels are listed in [Model Market](https://yir.ai/models).

Krea 3 edits one image or combines up to eight references. The order of the references in the request defines the numbered image tags in the prompt. The route returns one PNG at 1K or 2K and uses fal's medium reasoning.

fal bills output and processed reference pixels at $0.025 per 1024×1024 pixels, with each reference capped at the output pixel count. Read `primary_estimate` for the task budget based on those caps. fal Billing Events reports the final amount charged per task.
