Seedream 5.5 API for Images with Real Transparency

Seedream 5.5 is ByteDance's latest Seedream model for text-to-image and image editing. Send a prompt to POST /api/v1/images/generations with model: seedream-5-5, add up to 10 reference images in image_list for edits, and set background: transparent for a real alpha channel, with or without input images. Each request returns one image; 1K and 1.5K cost the same.
Get API Key
Glass product bottle floating above a clean studio surface with a soft shadow

Prompts, References and Alpha in One Model

Transparent output from a prompt

Set background: transparent to get a PNG with a real alpha channel. It works for pure text-to-image and with reference images, and the inputs do not need to be transparent. Combining it with output_format: jpeg is rejected.
Sheet of die-cut vinyl stickers peeled off a backing sheet

Let the model decide with auto

background: auto lets the model choose from your prompt and references. The result reports background as transparent or opaque so you know which you got. Output is PNG.
Two versions of the same illustrated mug, one on a checkerboard and one on a solid color

Up to 10 references for edits

Pass up to 10 images in image_list as [{"url": "https://…"}], each up to 30MB, to restyle, combine or edit. Each reference adds 0.42 tokens ($0.0084).
Mood board of fabric swatches and photos next to a finished fashion illustration
model and prompt are required; references and format options are optional.

Seedream 5.5 Request Fields and Limits

model

Always seedream-5-5. To split an image into layers use seedream-5-5-layers.

prompt (required)

Text describing the image or the edit, 1 to 4000 characters.

image_list

Optional, 1 to 10 reference images, each up to 30MB. Omit it for pure text-to-image.

ratio and image_size

ratio: auto (default), 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9 or 9:21. image_size: 1K, 1.5K or 2K (default).

background

opaque (default), transparent or auto. transparent and auto output PNG and cannot be combined with output_format: jpeg.

output_format

jpeg (default) or png; defaults to png when background is transparent or auto.
An output price by tier plus a small fee per reference image.

Seedream 5.5 API Pricing

1.04 tokens ($0.0208) at 1K or 1.5K

1K and 1.5K share one price, so 1.5K gives more pixels for the same cost.

1.82 tokens ($0.0364) at 2K

2K is the default tier.

0.42 tokens ($0.0084) per reference image

Each image in image_list adds this fee. Example: 2K with 3 references costs 3.08 tokens ($0.0616).

Refunded on failure

Tokens are deducted when the task is accepted and returned automatically if it fails; cost in the response shows the amount actually charged.

Integrate the Seedream 5.5 API

1

1. Create an API key

Generate a key at /keys and send it as Authorization: Bearer <key>.
2

2. Build the request

Start from {"model": "seedream-5-5", "prompt": "…", "ratio": "4:5"}; add image_list for edits or background: transparent for cut-out assets.
3

3. Submit and store

POST to /api/v1/images/generations and save the returned img_uuid and cost.
4

4. Fetch the image

Poll GET /api/v1/jobs/detail or use custom_callback_url; download the image before link_expired_at, 24 hours after completion.
Create an API Key
Image features that need clean assets and reference control.

What Developers Build with Seedream 5.5

Product cut-outs

Generate product shots on a transparent background, ready to drop onto any page or banner.

Stickers, icons and overlays

Create PNG assets with real alpha straight from a prompt, no background removal step.

Reference-guided edits

Combine up to 10 references to keep a subject, style or layout consistent across variations.

Ad and thumbnail variations

Produce many prompt variations at 1.5K for 1.04 tokens ($0.0208) each.
Transparency, references and pricing.

Seedream 5.5 API FAQ

What is the Seedream 5.5 API endpoint?

POST https://api.goenhance.ai/api/v1/images/generations with model: seedream-5-5.

Does transparent output need a transparent input?

No. background: transparent works for pure text-to-image and with any reference images.

What does background auto do?

The model decides from your prompt and references, and the result reports background as transparent or opaque.

How many images does a request return?

Always a single image.

How are reference images billed?

Each image in image_list adds 0.42 tokens ($0.0084), up to 10 images.

How do I get editable layers?

Use the separate seedream-5-5-layers model, which splits one image into a base plus up to 99 transparent layers.

Start Building with the Seedream 5.5 API

Seedream 5.5 API: Pricing & Integration | GoEnhance