ByteDance
Seedream 5.5
Build a composition from up to ten references, generate transparent artwork from text, or split a design into layers. Text readback makes it easier to locate lettering for a follow-up edit.
bytedance/seedream-5-5Best for
- Generate ready-to-place transparent campaign artwork
- Separate design layers and recover their lettering
Technical details
Read the model documentation- API model ID
bytedance/seedream-5-5- Provider
- ByteDance
- Modality
- Image
- Input
- Text prompt · Images
- Output
- Image
- Endpoint
POST /v1/byteplus/images/async- Response formats
- url
Request parameters
promptstringWhat to generate. Optional in layer-decomposition mode, where the model finds the elements itself; required otherwise. Requirement: yes*.
text
sizestring · Optional · Default: 2K (auto when decomposing)Choose a size preset and describe the aspect ratio in the prompt, or give an exact width and height.
1K · 1.5K · 2K, or WxH in pixels
- Custom sizes allow 921,600 to 4,624,220 pixels in total and aspect ratios between 1:16 and 16:1.
- Decomposition accepts presets and auto.
output_formatstring · Optional · Default: pngIn decomposition mode this governs the base image only — layers are always png.
- png
- jpeg
response_formatstring · Optional · Default: urlThe async route is url-only; b64_json is refused, because buffering full images would sink the relay.
url
backgroundstring · Optional · Default: autotransparent returns a live alpha channel, from a prompt alone or from reference images.
- auto
- transparent
- opaque
- Not with layer decomposition (an error), and the output must be png: output_format: jpeg beside it is an error.
Additional parameters
imagestring or string[] · OptionalPresent -> image-to-image, up to 10 reference images.
public https urls, or inline data:image/<fmt>;base64,…
- Inline images are taken on POST /v1/byteplus/images/async only, within its 256 KiB body cap, so a url is the safer choice.
- Exactly ONE png or jpeg in layer-decomposition mode;
- more is an error.
layer_decompositionboolean · Optional · Default: falseTakes one image apart into a base image plus layers, up to 100 images in all.
- true
- false
- The poll then returns every one of them in data[].
- Every image is charged at its own size — a 30-image result is 30 charges, not one.
layer_decomposition_optionsobject · OptionalOnly with layer_decomposition: true.
max_images: 2-100 · text_recognition: true · false
- max_images caps the result, base image included (default 30).
- text_recognition (default false) returns, for each regular-text layer, its text, its lines with their boxes, and the style of each run.
optimize_prompt_optionsobject · Optional{"mode": "standard"} — the only mode this model offers.
mode: standard
safety_identifierstring · OptionalA stable ID for the end user. Sociaro replaces it with an organization-specific pseudonym before sending the request.
1–64 characters
- The same ID maps to the same pseudonym across your organization’s keys and routes. The response returns that pseudonym.
watermarkboolean · Optional · Default: the vendor's (on)BytePlus defaults this ON, and this route forwards your body as sent — send false for an image without the mark.
- true
- false
Pricing
| Billable unit and options | Rate (USD) |
|---|---|
| Per imageup to 2.6 MP (1K, 1.5K) | $0.022 |
| Per imageover 2.6 MP (2K) | $0.0385 |
| Per input image | $0.0088 |
Billing notes
- In plain generation, leaving out size gives 2K, which bills at the higher tier.
- With layer decomposition, leaving it out gives auto: each image is sized from the input image and bills at the tier its own size falls in.
- Layer decomposition is charged at the same rates: every output image, the base and each layer, by its own size; a layer smaller than 1K is charged at the lower tier.
Catalogue rates on 21 Nov 2026. Prices exclude VAT.
View in pricing catalogue