> ## Documentation Index
> Fetch the complete documentation index at: https://docs.qingbo.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Seedream Series

> ByteDance Doubao image generation — multi-version evolution, multi-image reference / image editing / image group generation

ByteDance Seed team's image generation series, with **three generations**:

* **5.0 Lite** — Visual reasoning (CoT) + real-time web search, 14 reference images, 2K/3K, strong identity preservation, cost-effective tier
* **4.5** — Professional production grade, native 4K, 14 reference images, 94% typography accuracy, 90% cross-angle character consistency
* **4.0** — Multimodal creation, 2-10 reference image fusion, unified interface for text-to-image / editing / image groups

Per-image pricing, generates 1K-4K images (range varies by version).

## Pricing

| Model                      | Price (per image) |
| -------------------------- | ----------------- |
| `doubao-seedream-5.0-lite` | `$0.02975`        |
| `doubao-seedream-4.5`      | `$0.02975`        |
| `doubao-seedream-4.0`      | `$0.0238`         |

## Mode Quick Reference

| Mode                                           | Trigger Field                              | Supported Versions                    |
| ---------------------------------------------- | ------------------------------------------ | ------------------------------------- |
| **Text-to-image**                              | `prompt` only                              | All                                   |
| **Image edit (inpainting / lighting changes)** | `+ image_urls` (single) + `action: "edit"` | 4.0 / 4.5                             |
| **Multi-image reference fusion**               | `+ image_urls` (multiple)                  | All (4.0=2-10 / 4.5=14 / 5.0-lite=14) |
| **Image group generation (themed set)**        | `sequential_image_generation: "auto"`      | All                                   |

## Examples

<CodeGroup>
  ```bash Text-to-image (minimal) theme={"system"}
  curl -X POST https://www.qingbo.dev/v1/tasks \
    -H "Authorization: Bearer $WAVE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "doubao-seedream-5.0-lite",
      "prompt": "Futuristic city night view, neon lights reflected on wet streets, cinematic composition",
      "resolution": "2K",
      "aspect_ratio": "16:9"
    }'
  ```

  ```bash Multi-image reference fusion (4.5) theme={"system"}
  curl -X POST https://www.qingbo.dev/v1/tasks \
    -H "Authorization: Bearer $WAVE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "doubao-seedream-4.5",
      "prompt": "Place the character from the first image into the scene from the second image",
      "image_urls": [
        "https://cdn.example.com/character.jpg",
        "https://cdn.example.com/scene.jpg"
      ],
      "resolution": "4K",
      "aspect_ratio": "16:9"
    }'
  ```

  ```bash Image edit (4.0) theme={"system"}
  curl -X POST https://www.qingbo.dev/v1/tasks \
    -H "Authorization: Bearer $WAVE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "doubao-seedream-4.0",
      "action": "edit",
      "prompt": "Replace the background with a sunset sky, keep the subject unchanged",
      "image_urls": ["https://cdn.example.com/portrait.jpg"],
      "resolution": "2K"
    }'
  ```

  ```bash Image group (themed set) theme={"system"}
  curl -X POST https://www.qingbo.dev/v1/tasks \
    -H "Authorization: Bearer $WAVE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "doubao-seedream-4.5",
      "prompt": "Four e-commerce hero images: front / side / detail / lifestyle",
      "sequential_image_generation": "auto",
      "sequential_image_generation_options": {"max_images": 4},
      "resolution": "4K",
      "aspect_ratio": "1:1"
    }'
  ```
</CodeGroup>

<ResponseExample>
  ```json Submission successful theme={"system"}
  {
    "task_id": "task-wave1775285160b950328499",
    "model": "doubao-seedream-4.5",
    "action": "generate",
    "status": "queued",
    "created_at": 1775285160040,
    "progress": 0
  }
  ```
</ResponseExample>

After submission, poll status with [`GET /v1/tasks/{task_id}`](/en/api-reference/task/status). See [Task System](/en/docs/task-system) for details.

## Available Models

| Model ID                   | Description                                                                                              |
| -------------------------- | -------------------------------------------------------------------------------------------------------- |
| `doubao-seedream-5.0-lite` | Visual reasoning + web search, 14 reference images, 2K/3K, strong identity preservation                  |
| `doubao-seedream-4.5`      | Professional production grade, 4K, 14 reference images, best typography / character consistency          |
| `doubao-seedream-4.0`      | Multimodal creation, 2-10 reference images, unified interface for text-to-image / editing / image groups |

## Common Parameters

<ParamField body="model" type="string" required>
  Choose one from the [Available Models](#available-models) list
</ParamField>

<ParamField body="action" type="string" default="generate">
  Operation type. Options:

  * `generate` — Text-to-image (default)
  * `image2image` — Image-to-image (use with `image_urls`)
  * `edit` — Image editing (inpainting / lighting changes, **4.0 / 4.5 only**)
  * `reference` — Multi-image reference fusion (**4.0 / 4.5 only**)
  * `group` — Image group generation (all versions)
</ParamField>

<ParamField body="prompt" type="string" required>
  Image description. Supports Chinese and English. Seedream achieves \~94% accuracy on complex typography and small text rendering
</ParamField>

<ParamField body="n" type="integer" default="1">
  Number of images to generate (returned in a single call)
</ParamField>

<ParamField body="seed" type="integer" default="-1">
  Random seed. `-1` for random; a fixed value reproduces similar results
</ParamField>

<ParamField body="aspect_ratio" type="string" default="1:1">
  Aspect ratio. Options:

  * `1:1` — Square
  * `4:3` / `3:4` — Standard landscape / portrait
  * `16:9` / `9:16` — Widescreen landscape / portrait
  * `3:2` / `2:3` — Camera ratio landscape / portrait
  * `21:9` — Ultra-wide (all versions)
  * `9:21` — Ultra-tall (4.0 / 4.5 only)
  * `auto` — Smart selection (follows reference image ratio)
</ParamField>

<ParamField body="resolution" type="string" default="2K">
  Output resolution, **depends on the model**:

  * 4.0: `1K` / `2K` / `4K`
  * 4.5: `2K` / `4K`
  * 5.0-lite: `2K` / `3K`
</ParamField>

<ParamField body="image_urls" type="string[]">
  Array of reference image URLs. **Count depends on the model**:

  * 4.0: 2-10 images (combined with final output ≤15)
  * 4.5: up to 14
  * 5.0-lite: up to 14
</ParamField>

<ParamField body="callback_url" type="string">
  Webhook callback URL, invoked when the task reaches a terminal state. See [Callback Mechanism](/en/docs/sync-async#async-task-webhook)
</ParamField>

## Model-Specific Parameters

<Tabs>
  <Tab title="Seedream 5.0 Lite">
    **Supported actions**: `generate` / `image2image` / `group` (note: **does not support `edit` or `reference`**)

    **resolution**: `2K` / `3K`

    **Key features**: Visual reasoning (CoT) + real-time web search, can reason from up-to-date world knowledge; strong identity preservation (eyes / jawline / proportions / skin tone preserved across styles)

    <ParamField body="watermark" type="boolean" default="false">
      AI-generated watermark
    </ParamField>

    <ParamField body="output_format" type="string" default="jpeg">
      **5.0-lite only** — Output format. Options:

      * `jpeg`
      * `png`
    </ParamField>

    <ParamField body="sequential_image_generation" type="string" default="disabled">
      Image group generation mode. Options:

      * `disabled` — Single image (default)
      * `auto` — Auto group (multiple themed images)
    </ParamField>

    <ParamField body="sequential_image_generation_options" type="object">
      Group options, e.g. `{"max_images": 4}` (range 1-15)
    </ParamField>
  </Tab>

  <Tab title="Seedream 4.5">
    **Supported actions**: `generate` / `image2image` / `edit` / `reference` / `group`

    **resolution**: `2K` / `4K`

    **Key features**: Native 4K, 14 reference images, \~90% cross-angle character consistency, \~94% typography and dense-text rendering accuracy

    <ParamField body="watermark" type="boolean" default="false">
      AI-generated watermark
    </ParamField>

    <ParamField body="optimize_prompt_options" type="object">
      Prompt optimization (nested object) with a `mode` field. Example: `{"mode": "standard"}`
    </ParamField>

    <ParamField body="sequential_image_generation" type="string" default="disabled">
      Group mode. Options:

      * `disabled` — Single image
      * `auto` — Auto group
    </ParamField>

    <ParamField body="sequential_image_generation_options" type="object">
      Group options, e.g. `{"max_images": 4}` (range 1-15)
    </ParamField>
  </Tab>

  <Tab title="Seedream 4.0">
    **Supported actions**: `generate` / `image2image` / `edit` / `reference` / `group`

    **resolution**: `1K` / `2K` / `4K`

    **Key features**: Origin of multimodal creation; unified interface for text-to-image / editing / image groups; 2-10 reference images, combined with final output ≤15

    <ParamField body="watermark" type="boolean" default="false">
      AI-generated watermark
    </ParamField>

    <ParamField body="optimize_prompt_options" type="string" default="standard">
      Prompt optimization (string, unlike 4.5's nested object). Options:

      * `standard` — Standard optimization (default)
      * `fast` — Fast mode
    </ParamField>

    <ParamField body="sequential_image_generation" type="string" default="disabled">
      Group mode. Options:

      * `disabled` — Single image
      * `auto` — Auto group
    </ParamField>

    <ParamField body="sequential_image_generation_options" type="object">
      Group options, e.g. `{"max_images": 4}` (range 1-15)
    </ParamField>
  </Tab>
</Tabs>

## Resource Limits

| Item                              | Limit                                                           |
| --------------------------------- | --------------------------------------------------------------- |
| Reference images (4.0)            | 2-10, combined with final output ≤15                            |
| Reference images (4.5 / 5.0-lite) | Up to 14                                                        |
| Single image size                 | ≤ 30MB, supports JPG/PNG/WEBP                                   |
| Output resolution                 | See each tab (up to 4K)                                         |
| Output file                       | JPG (default) / PNG (5.0-lite optional), URL valid for 24 hours |

## Related Docs

* [Task System](/en/docs/task-system) — Task state machine / polling cadence / async push
* [Request & Response](/en/docs/request-response) — Common error codes / Headers / rate limits
* [Authentication](/en/docs/authentication) — API Key application and usage