NOV 9, 2026
Table of contents
Seedream 5.5 and transparent backgrounds
November 9, 2026
Dreamina added ByteDance’s Seedream 5.5 to Canadian accounts, and the Dreamina API supports it now on POST /images:
model: "seedream-5.5"makes 4 images per request at 1.5K (default) or 2K, from text and up to 6 reference images, and picks the aspect ratio for you when you don’t set one- On a Canadian account it uses 8 credits an image at 1.5K and 13 at 2K (launch price, normally 11 and 18)
- The new
backgroundparameter takesauto(default),opaqueortransparent. Withtransparentyou get PNG images with a transparent background. It works withseedream-5.5,gpt-image-2,gpt-image-2.5-flareandgpt-image-2.5-sunburst
Below are the real outputs of our test round, all made through the API on a Canadian account.
Seedream 5.5 at 1.5K
One request, four images of 1536×1536, ready in 23 seconds.

curl — POST /images
curl --location 'https://api.useapi.net/v1/dreamina/images' \
--header 'Authorization: Bearer user:12345-…' \
--form 'model="seedream-5.5"' \
--form 'resolution="1.5k"' \
--form 'ratio="1:1"' \
--form 'prompt="a red vintage bicycle leaning on a white brick wall, soft morning light, photo"'
Transparent background
The same prompt with background: "transparent". Seedream 5.5 returned four PNGs with an alpha channel, shown here on a checkerboard so you can see what is transparent. Because the prompt describes a scene, Seedream kept part of the wall in three of them and the whole scene in the fourth.

curl
curl --location 'https://api.useapi.net/v1/dreamina/images' \
--header 'Authorization: Bearer user:12345-…' \
--form 'model="seedream-5.5"' \
--form 'resolution="1.5k"' \
--form 'ratio="1:1"' \
--form 'background="transparent"' \
--form 'prompt="a red vintage bicycle leaning on a white brick wall, soft morning light, photo"'
GPT Image 2.5 Flare, given the same prompt and background: "transparent", dropped the wall and returned a clean cutout of the bicycle alone, one 1328×1328 PNG in 33 seconds for 5 credits. Download the original PNG.

curl
curl --location 'https://api.useapi.net/v1/dreamina/images' \
--header 'Authorization: Bearer user:12345-…' \
--form 'model="gpt-image-2.5-flare"' \
--form 'resolution="1k"' \
--form 'ratio="1:1"' \
--form 'background="transparent"' \
--form 'prompt="a red vintage bicycle leaning on a white brick wall, soft morning light, photo"'
For stickers, product shots and logos, describe the object on its own (“a red vintage bicycle, isolated”) rather than a scene, and both models have less background to decide about.
Also new
- US accounts now run Seedream 5.0 Pro (
seedream-5.0-pro, 2K or 4K) and Seedream 5.0 Flash (seedream-5.0-flash, 1.5K, 2K or 4K, 6 credits an image up to 2K). Flash replaced Seedream 5.0 Lite on US, which stays on Canadian accounts - Canadian accounts get Seedream 3.1 (
seedream-3.1, text only, 1K or 2K) - Dreamina dropped Nano Banana and Seedance 1.5 Pro from Canadian accounts. Seedance 1.5 Pro still runs on US accounts
Full parameter reference: POST /images. Credits per image for every model: Setup Dreamina.