# Seedream 4.0

Seedream 4.0 pairs multimodal input with multi-image output. One call can take a text prompt, reference images, or both, and return a coherent set of high-resolution images.

- **Model ID:** `bytedance/seedream-4.0`
- **Type:** image
- **Providers:** bytedance
- **Context window:** Not applicable
- **Maximum output tokens:** 0
- **Pricing:** $0.03/image
- **Canonical page:** https://vercel.com/ai-gateway/models/seedream-4.0

## Supported parameters

- **Parameters:** `n`
- **Operations:** `text-to-image`, `image-to-image`
- **Input modalities:** `text`, `image`
- **Output modalities:** `image`
- **Aspect ratios:** Not reported
- **Resolutions:** Not reported
- **Durations:** Not reported
- **Frame rates:** Not reported
- **Maximum outputs per request:** 9
- **Maximum input images:** 6
- **Input image formats:** `jpeg`, `png`
- **Maximum input image size:** 10 MB

## Example

```ts
import { experimental_generateImage as generateImage } from 'ai';

const result = await generateImage({
  model: 'bytedance/seedream-4.0',
  prompt: 'A red balloon on a wooden table.'
});
```

## About

Seedream 4.0 was released September 9, 2025 as a multimodal image creation model from ByteDance. It extends beyond text-to-image generation by accepting text, single-image, and multi-image inputs natively in a single endpoint, then returning either one image or a set of related images with consistent subjects.

Multi-image input opens workflows that earlier text-to-image models couldn't cover in one call. You can blend several reference images, edit a source photograph with prompt-driven changes, or generate a series of frames that hold a subject's identity across scenes. ByteDance positions this as the foundation for catalog generation, batch editing, and reference-consistent series production.

Output resolution reaches up to 4K, which suits print layouts, hero imagery, and high-resolution editorial use without a separate upscaling step. ByteDance reports a per-image generation time near 1.8 seconds in published examples, a meaningful step up from earlier Seedream releases.

Seedream 4.0 is a dedicated image model. Call it through the AI SDK's `generateImage` function with the model set to `bytedance/seedream-4.0`. AI Gateway handles authentication and provider routing.

## What to consider

Image-generation pricing on AI Gateway is per image, listed at $0.03 for Seedream 4.0. Resolution and aspect ratio choices affect downstream usage but not the per-image rate. Review N/A when planning batch workflows.

## When to use

### Best For

- Multi-image composition workflows where several reference images must blend into a single coherent output
- Catalog generation and batch editing pipelines that need subject consistency across many related images
- High-resolution hero imagery, print layouts, and editorial use where 4K output removes a separate upscaling step
- Interactive design tools where roughly 1.8-second generation keeps the iteration loop short

### Consider Alternatives When

- You need video output rather than a still image. Seedance 2.0 or Seedance 2.0 Fast handle text-to-video and image-to-video
- Editing consistency on portraits, small text, and color tone is the primary requirement. Seedream 4.5 raises that ceiling
- The workflow depends on web-connected retrieval grounded in real-time information. Seedream 5.0 Lite adds that capability

## Best for

- **Multi-Image Composition:** Blend several reference images into a single coherent output in one call
- **Catalog and Batch Generation:** Produce many related images that hold subject identity across the set
- **High-Resolution Output:** Up to 4K resolution removes a separate upscaling step for print and editorial use
- **Fast Iteration Loops:** Roughly 1.8-second generation suits interactive design tools and prompt exploration

## Consider alternatives

- **Video Output:** Seedance 2.0 or Seedance 2.0 Fast cover text-to-video and image-to-video workflows
- **Edit-Consistency Workloads:** Seedream 4.5 improves portrait detail, small-text rendering, and color preservation
- **Web-Grounded Generation:** Seedream 5.0 Lite adds web-connected retrieval for real-time context

## Frequently asked questions

### What input types does Seedream 4.0 support?

Text prompts, a single reference image, or multiple reference images in the same request. You can combine modalities to guide composition and subject consistency in one call.

### Can Seedream 4.0 return more than one image per request?

Yes. Seedream 4.0 can output a single image or a set of related images that maintain subject consistency, which suits catalog generation and frame-series workflows.

### What output resolution does Seedream 4.0 support?

Up to 4K. That removes a separate upscaling step for print layouts, hero imagery, and editorial use.

### How fast is Seedream 4.0 per image?

ByteDance reports generation around 1.8 seconds per image in published examples. See live metrics on this page for current throughput across providers.

### How do I call Seedream 4.0 through AI Gateway?

Set the model to `bytedance/seedream-4.0` and call it through the AI SDK's `generateImage` function. AI Gateway routes the request through `bytedance` using your API key or OIDC token.

### What does Seedream 4.0 cost per image?

Seedream 4.0 is listed at $0.03 per image. See N/A for the full pricing breakdown.

### Does Vercel AI Gateway support Zero Data Retention for Seedream 4.0?

Zero Data Retention is not currently available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/seedream-4.0)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://docs.byteplus.com/en/docs)
- [Provider pricing](https://docs.byteplus.com/en/docs/ModelArk/1544106)


If you encounter incorrect information, can’t find what you need, need a capability Vercel doesn’t currently offer, or have any other feedback to report, please do it by following the instructions at https://vercel.com/feedback.md.