ByteDance Doubao Seedream 5.0 Pro flagship image API. Highlights: interactive editing (mark exactly what to change with coordinates / box-select / arrows / hand-drawn marks — edits the region, keeps the rest), multilingual in-image text rendering (14+ languages) and multi-image fusion (up to 10 references). 1K/2K single-image output, no watermark.
All requests carry the API key in the header:
Authorization: Bearer YOUR_API_KEY| Field | Required | Type | Description |
|---|---|---|---|
| Resolution | Price / image | ||
| 1K | $0.03 | ||
| 2K | $0.06 |
Billed per successfully generated image (single image per call); failed requests are not charged. No image series or 4K.
/api/v1/images/generationsAsync contract: create returns a taskId; poll GET ?task_id= until completed, or receive a webhook via callback_url.
# Text-to-image — create returns a taskId, then poll for the result.
curl -X POST https://api.apimodels.app/v1/images/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedream-5-0-pro",
"prompt": "a travel magazine cover, bold title \"SEEDREAM\", subtitle \"EXPLORE THE WILD\", a scarlet macaw in a sunlit rainforest, cinematic",
"resolution": "2K",
"output_format": "jpeg"
}'
# → { "code": 200, "data": { "taskId": "...", "state": "pending", "model": "doubao-seedream-5-0-pro" } }
curl "https://api.apimodels.app/v1/images/generations?task_id=TASK_ID" \
-H "Authorization: Bearer YOUR_API_KEY"
# → { "code": 200, "data": { "state": "completed", "resultUrls": ["https://r2.apimodels.app/..."] } }Same endpoint: include image (single) or images[] (up to 10) to run image-to-image / editing / multi-image fusion. Interactive editing is driven by the prompt (mark the edit region on the reference with coordinates / box-select / arrows / hand-drawn marks).
# Image editing / multi-image fusion — pass one or more reference images.
# Interactive editing (coordinates / box / arrows / hand-drawn marks) is driven
# by the PROMPT on the reference image.
curl -X POST https://api.apimodels.app/v1/images/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedream-5-0-pro",
"prompt": "combine the product from image 1 with the scene in image 2, clean magazine layout",
"images": ["https://example.com/product.png", "https://example.com/scene.jpg"],
"resolution": "2K"
}'
# Single-image edit — one reference + an instruction
curl -X POST https://api.apimodels.app/v1/images/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedream-5-0-pro",
"prompt": "keep the composition; add a ceramic coffee cup in the lower-right, make it blend naturally",
"image": "https://example.com/desk.jpg",
"resolution": "2K"
}'| Field | Required | Type | Description |
|---|---|---|---|
| Field | Type | Description | |
| model | string | doubao-seedream-5-0-pro | |
| prompt | string | Prompt (required); editing instructions and coordinate marks go here | |
| resolution | string | 1K or 2K (default 2K) | |
| aspect_ratio | string | 1:1 / 4:3 / 3:4 / 16:9 / 9:16 / 3:2 / 2:3 / 5:4 / 4:5 / 21:9 (square when omitted) | |
| image / image_url / image_base64 | string | Single reference image → image-to-image / edit | |
| images | string[] | Multiple references, up to 10 (URL or data URI) | |
| output_format | string | jpeg (default) or png | |
| callback_url | string | Webhook URL called when the task completes |
Same endpoint and request fields — just set model to seedream-5.0-pro-spicy. Same image quality as Pro; the difference is looser content moderation (we add no filter of our own, but the model's own safety systems still apply upstream). 1K / 2K only (a 4K request returns 400 and is not charged), JPEG output, typically 40–80 seconds per image.
| Field | Required | Type | Description |
|---|---|---|---|
| Resolution | Price / image | ||
| 1K | $0.035 | ||
| 2K | $0.06 (default) |
Charged only on success. Try it in the Playground.
Try it in the Playground. For 2K volume runs (Flash charges the same for 1K and 2K — half of Pro at 2K) or layer splitting (not available on Pro), see Seedream 5.0 Flash ($0.03/image).
Pro adds interactive editing (mark the exact region to change via coordinates / box-select / arrows / hand-drawn marks) and stronger multilingual in-image text rendering. It outputs a single 1K or 2K image (no series, no 4K).
$0.03 per image at 1K and $0.06 at 2K, billed only on success. One API key on apimodels.app, more affordable than going direct.
Up to 10 references for multi-image fusion / editing — pass them as images[] (or a single image). Image-to-image whose output still carries a real person is rejected by content review.