ByteDance

Seedream 4.0

Name: Seedream 4.0
Brand: ByteDance
Price: 0.03 USD
Availability: InStock

AI Image Editing Model

Mode:

Image

ByteDance's image generation model with integrated text-to-image and image editing capabilities in a unified architecture, supporting up to 4K resolution

Seedream 4.0 generated image of A hyper-realistic close-up of an artisan watchmaker's workbench, shallow dept...

Try Model Benchmark examples

4096 x 4096

Max Resolution

Supported Modes

Text to Image Image Edit

Active

Details

Model ID

seedream-4

Also known as: seedream-4.0

Creator

ByteDance

Family

seedream

Released

September 2025

Ready to integrate?

Access seedream-4 via our unified API.

Create Account

Available at 3 providers

Starting from

$0.030 /image via BytePlus, fal.ai, Replicate

Prices shown are in USD

See all providers

Providers & Pricing (3)

Seedream 4.0 is available from 3 providers, with per-image pricing starting at $0.03 through BytePlus.

BytePlus

Text to Image Image Edit

byteplus/seedream-4

Provider Model ID: seedream-4-0

$0.030 /image

View pricing details

fal.ai

Image Edit

fal/seedream-4-edit

Provider Model ID: fal-ai/bytedance/seedream/v4/edit

$0.030 /image

View pricing details

Replicate

Text to Image Image Edit

replicate/seedream-4

Provider Model ID: bytedance/seedream-4

$0.030 /image

View pricing details

seedream-4.0 API OpenAI-compatible

Access Seedream 4.0 via Lumenfall’s OpenAI-compatible API to generate 4K images and perform complex image edits through a single, unified endpoint. Developers can integrate ByteDance's text-to-image and editing architecture into any application using standard API protocols.

Base URL

https://api.lumenfall.ai/openai/v1

Model

seedream-4

Code Examples

Image Edit

/v1/images/edits

curl -X POST \

  https://api.lumenfall.ai/openai/v1/images/edits \

  -H "Authorization: Bearer $LUMENFALL_API_KEY" \

  -F "model=seedream-4" \

  -F "[email protected]" \

  -F "prompt=Add a starry night sky to this image" \

  -F "size=1024x1024"

# Response:

# { "created": 1234567890, "data": [{ "url": "https://...", "revised_prompt": "..." }] }

import OpenAI from 'openai';

import fs from 'fs';

const client = new OpenAI({

  apiKey: 'YOUR_API_KEY',

  baseURL: 'https://api.lumenfall.ai/openai/v1'

});

const response = await client.images.edit({

  model: 'seedream-4',

  image: fs.createReadStream('source.png'),

  prompt: 'Add a starry night sky to this image',

  size: '1024x1024'

});

// { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }

console.log(response.data[0].url);

from openai import OpenAI

client = OpenAI(

    api_key="YOUR_API_KEY",

    base_url="https://api.lumenfall.ai/openai/v1"

response = client.images.edit(

    model="seedream-4",

    image=open("source.png", "rb"),

    prompt="Add a starry night sky to this image",

    size="1024x1024"

# { created: 1234567890, data: [{ url: "https://...", revised_prompt: "..." }] }

print(response.data[0].url)

Parameter Reference

Required Supported Not available

Core Parameters

Parameter	Type	Description	Modes
`prompt`	string	Required. Text prompt for image generation	T2I Edit
`seed`	integer	Random seed for reproducibility	T2I Edit

Size & Layout

Parameter	Type	Description	Modes
`size`	string	Image dimensions as WxH pixels (e.g. "1024x1024") or aspect ratio (e.g. "16:9") WxH determines both shape and scale (aspect_ratio and resolution are ignored when size is provided). W:H format is equivalent to aspect_ratio.	T2I Edit
`aspect_ratio`	string	Aspect ratio of the output image (e.g. "16:9", "1:1") Controls shape independently of scale. Use with resolution to control both. If size is also provided, size takes precedence. Any ratio is accepted and mapped to the nearest supported value.	T2I Edit
`resolution`	string	Output resolution tier (e.g. "1K", "4K") `auto` `1K` Controls scale independently of shape. Higher tiers produce larger images and cost more. If size is also provided, size takes precedence for scale. Any tier is accepted and mapped to the nearest supported value.	T2I Edit

Output	`size`		`aspect_ratio` + `resolution`
Flexible
Auto	`"auto"`	—	Model chooses optimal dimensions
Custom 1–14142px per side	`"WxH"`	—	Any pixel dimensions within model constraints

1K 8 sizes

Output	`size`		`aspect_ratio` + `resolution`
1183 × 887	`"1183x887"`	or	`"4:3"` + `"1K"`
1024 × 1024	`"1024x1024"`	or	`"1:1"` + `"1K"`
887 × 1182	`"887x1182"`	or	`"3:4"` + `"1K"`
836 × 1254	`"836x1254"`	or	`"2:3"` + `"1K"`
1254 × 836	`"1254x836"`	or	`"3:2"` + `"1K"`
768 × 1365	`"768x1365"`	or	`"9:16"` + `"1K"`
1365 × 768	`"1365x768"`	or	`"16:9"` + `"1K"`
1563 × 670	`"1563x670"`	or	`"21:9"` + `"1K"`

How these parameters work

size

Exact pixel dimensions

"1920x1080"

aspect_ratio

Shape only, default scale

"16:9"

resolution

Scale tier, preserves shape

"1K"

Priority when combined

size › aspect_ratio + resolution › aspect_ratio › resolution

size is most specific and always wins. aspect_ratio and resolution control shape and scale independently.

How matching works

Shape matching – we pick the closest supported ratio. Ask for 7:1 on a model with 4:1 and 8:1, you get 8:1.

Scale matching – providers use different tier formats: K tiers (0.5K 1K 2K 4K) or megapixel tiers (0.25 1). If the exact tier isn't available, you get the nearest one.

Dimension clamping – if a model has pixel limits, we clamp dimensions to fit and keep the aspect ratio intact.

Media Inputs

Parameter	Type	Description	Modes
`image`	file	Required. Input image(s) to edit Supports PNG, JPEG, WebP.	T2I Edit

Output & Format

Parameter	Type	Description	Modes
`response_format` byteplus	string	How generated images are returned. `b64_json` `url` Default: `"url"`	T2I Edit
`output_format`	string	Output image format `png` `jpeg` `gif` `webp` `avif` Gateway converts to requested format if provider doesn't support it natively.	T2I Edit
`output_compression`	integer	Compression level for lossy formats (JPEG, WebP, AVIF)	T2I Edit
`n`	integer	Number of images to generate Default: `1` Gateway generates multiple images in parallel even if provider only supports 1.	T2I Edit

Additional Parameters

Provider-specific passthrough fields, available only when the request is routed to the listed provider.

Parameter	Type	Description	Modes
Universal
`prompt_enhancement`	string	Whether an LLM rewrites/expands the prompt before generation (off, on) `off` `on`	T2I Edit
`sequential_image_generation`	string	Group image generation mode. 'disabled' generates a single image. 'auto' lets the model decide whether to generate multiple related images (e.g., story scenes, character variations). `disabled`	T2I Edit
byteplus
`sequential_image_generation_options`	object	Batch image generation options, effective only when sequential_image_generation is auto.	T2I Edit
`stream`	boolean	Whether to stream each output image as soon as it is generated.	T2I Edit
`watermark`	boolean	Whether to add an AI-generated watermark to the lower-right corner of the image.	T2I Edit
fal
`enable_safety_checker`	boolean	If set to true, the safety checker will be enabled.	T2I Edit
`sync_mode`	boolean	If `True`, the media will be returned as a data URI and the output data won't be available in the request history.	T2I Edit
replicate
`height`	integer	Custom image height (only used when size='custom'). Range: 1024-4096 pixels.	T2I Edit
`width`	integer	Custom image width (only used when size='custom'). Range: 1024-4096 pixels.	T2I Edit

Parameter Normalization

How we handle parameters across different providers

Not every provider speaks the same language. When you send a parameter, we handle it in one of four ways depending on what the model supports:

Behavior	What happens	Example
`passthrough`	Sent as-is to the provider	style, quality
`renamed`	Same value, mapped to the field name the provider expects	prompt
`converted`	Transformed to the provider's native format	size
`emulated`	Works even if the provider has no concept of it	n, response_format

Parameters we don't recognize pass straight through to the upstream API, so provider-specific options still work.

Full API Reference Authentication, endpoints, and more

Seedream 4.0 Benchmarks

Seedream 4.0 currently holds the #17 position for text-to-image generation with a 1237 Elo rating and ranks #14 for image editing with an Elo of 1174. This unified architecture from ByteDance maintains competitive performance across both creative generation and precise image manipulation tasks.

Lumenfall Arena

#16

Text-to-Image

1244 Elo

Lumenfall Arena

#18

Image Editing

1170 Elo

Image Editing Landscape

Elo vs Cost

1 model without pricing omitted

Elo vs Speed

8 models waiting for enough speed data

Competition Results

Image Editing

Photorealism

View leaderboard

Seedream 4.0 edited result for Golden Hour Stroll

Original image before Seedream 4.0 editing

Before After

#13

Golden Hour Stroll

13 models

Image Editing

Edit instruction

“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”

#13

Man and Car in California

13 models

Image Editing

Source

Edit instruction

“Make a photo of the man driving the car down the California coastline”

Seedream 4.0 edited result for Bald man challenge

Before After

#15

Bald man challenge

15 models

Image Editing

Edit instruction

“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”

Seedream 4.0 edited result for Night Sky Transformation

Before After

Night Sky Transformation

16 models

Image Editing

Edit instruction

“Change the scene to night: a deep, dark sky with subtle, glistening stars visible behind the mountain.”

3 attempts – showing best result

Seedream 4.0 edited result for Neutral Expression to Genuine Smile

Before After

#13

Neutral Expression to Genuine Smile

14 models

Image Editing

Edit instruction

{
  "action": "image_edit",
  "reference": "uploaded neutral portrait",
  "change": "Warm genuine Duchenne smile: lips curved up, slight natural teeth, soft eye crinkles, subtle cheek raise",
  "details": "Realistic smiling skin (dimples if present, soft cheek shadows), slightly brighter eyes; keep exact eye shape/color/iris",
  "preserve_exact": "Face identity/structure, eyes/nose/lips/eyebrows, hair, skin texture/pores/freckles, makeup, clothing, head pose, background, lighting, shadows, framing",
  "no_changes": "No face shape change, no new features, no gaze shift, no hair/clothing/lighting/background edits",
  "style": "Ultra-photorealistic 8K portrait, sharp face focus, natural soft lighting, realistic skin glow"
}

4 attempts – showing best result

Image Editing

Anime

View leaderboard

#12

Studio Ghibli Anime Style

14 models

Image Editing

Source

Edit instruction

“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”

Image Editing

Portrait

View leaderboard

Before After

#15

Bald man challenge

15 models

Image Editing

Edit instruction

“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”

Before After

#13

Neutral Expression to Genuine Smile

14 models

Image Editing

Edit instruction

{
  "action": "image_edit",
  "reference": "uploaded neutral portrait",
  "change": "Warm genuine Duchenne smile: lips curved up, slight natural teeth, soft eye crinkles, subtle cheek raise",
  "details": "Realistic smiling skin (dimples if present, soft cheek shadows), slightly brighter eyes; keep exact eye shape/color/iris",
  "preserve_exact": "Face identity/structure, eyes/nose/lips/eyebrows, hair, skin texture/pores/freckles, makeup, clothing, head pose, background, lighting, shadows, framing",
  "no_changes": "No face shape change, no new features, no gaze shift, no hair/clothing/lighting/background edits",
  "style": "Ultra-photorealistic 8K portrait, sharp face focus, natural soft lighting, realistic skin glow"
}

4 attempts – showing best result

Uncategorized

Over-the-top cartoon caricature

13 models

Image Editing

Source

Edit instruction

“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”

Seedream 4.0 edited result for Selective Object Swap

Before After

Selective Object Swap

12 models

Image Editing

Edit instruction

“Swap the positions of the blue and yellow blocks”

2 attempts – showing best result

Top Matchups

See how Seedream 4.0 performs head-to-head against other AI models, ranked by community votes in blind comparisons.

vs FLUX.2 [max]

Challenge: Studio Ghibli Anime Style

67% W · 33% L

vs Grok Imagine Image

Challenge: Modern Clean Menu

0% W · 100% L

vs GPT Image 1.5

Challenge: Vintage Cafe Logo

33% W · 67% L

vs Nano Banana Pro

Challenge: Man and Car in California

0% W · 100% L

vs Nano Banana

Challenge: Bald man challenge

0% W · 100% L

Seedream 4.0 is best for

See all Use Cases

ByteDance's model excels in anime generation with a 71.4% win rate and ranks #6 for photorealism, though it shows relative weakness in portraiture and text rendering. It is a top-tier choice for commercial branding and photorealistic outputs where it maintains a 60% win rate against competing vision models.

#5 of 14

Anime

Edit · 61.8% win rate

Gallery

View all 8 images

Seedream 4.0 FAQ

How much does Seedream 4.0 cost?

Seedream 4.0 starts at $0.03 per image through Lumenfall. Pricing varies by provider. Lumenfall does not add any markup to provider pricing.

How do I use Seedream 4.0 via API?

You can use Seedream 4.0 through Lumenfall's OpenAI-compatible API. Send requests to the unified endpoint with model ID "seedream-4". Code examples are available in Python, JavaScript, and cURL.

Which providers offer Seedream 4.0?

Seedream 4.0 is available through Replicate, fal.ai, and BytePlus on Lumenfall. Lumenfall automatically routes requests to the best available provider.

What is the maximum resolution for Seedream 4.0?

Seedream 4.0 supports images up to 4096x4096 resolution.

Overview

Seedream 4.0 is a high-resolution image generation and editing model developed by ByteDance. It utilizes a unified architecture that handles both text-to-image synthesis and sophisticated image manipulation within the same framework. This version is specifically designed for high-fidelity outputs, supporting image generation at resolutions up to 4K.

Strengths

High-Resolution Output: Demonstrates superior performance in generating 4K imagery, maintaining sharpness and fine textures that often degrade in standard 1024px models.
Unified Task Processing: Unlike models that require separate adapters (like ControlNet) for editing, Seedream 4.0 integrates generation and editing capabilities into a single pipeline, improving consistency between original and edited pixels.
Multi-Modal Conditioning: Processes both text prompts and image inputs simultaneously, allowing for precise instruction-following when modifying existing compositions or styles.
Instruction Adherence: Shows high accuracy in translating complex descriptive prompts into visual elements, particularly in maintaining spatial relationships between objects.

Limitations

Inference Latency: Generating at 4K resolution requires significant computational resources, leading to slower generation times compared to lower-resolution models like SDXL or Flux Schnell.
Operational Cost: At a starting price of $0.03 per generation, it carries a higher cost-per-image than many base latent diffusion models, making it less ideal for high-volume batch processing of simple assets.
Prompt Sensitivity: While capable of high fidelity, the model’s unified architecture may occasionally require more specific prompting to distinguish between “generate new content” and “preserve existing content” during editing tasks.

Technical Background

Seedream 4.0 is built on an evolution of ByteDance’s diffusion research, moving away from fragmented architectures toward a unified sequence-to-sequence approach for visual tasks. It treats image editing and generation as a singular objective, which likely involves a high-capacity transformer-based backbone optimized for large-scale spatial parameters. The training regimen emphasizes high-resolution data scaling to ensure the model maintains coherence at 4K dimensions.

Best For

Seedream 4.0 is ideal for professional design workflows where high-print quality or large-format digital assets are required. It excels in scenarios involving “inpainting” and complex image-to-image transformations where the user needs to maintain the integrity of a base image while adding specific details.

Seedream 4.0 is available through Lumenfall’s unified API and playground, allowing developers to integrate its 4K generation and editing capabilities without managing complex infrastructure.

Try Seedream 4.0 in Playground

Generate images with custom prompts — no API key needed.

Open Playground