Inference Space Docs

Doubao Seedream 5.0 Pro text-to-image and multi-image-reference generation with up to 10 reference images, OpenAI compatible.

Doubao Seedream 5.0 Pro is well suited to product renders, illustration, and style-consistent asset generation, and supports text-to-image generation and multi-image-reference generation with up to 10 reference images.

The interface is compatible with the OpenAI Images API. For endpoints, how to pass reference images, common parameters, response format, and async jobs, see Image generation and editing; this page covers only the behavior specific to this model.

model: doubao-seedream-5-0-pro (the dated doubao-seedream-5-0-pro-260628 points to the same model at the same price).
Text-to-image: POST /v1/images/generations.
Multi-image reference: POST /v1/images/edits, with repeated multipart image[] fields or JSON images.

Text-to-image

# China region, accelerated route; China international route is https://global.inf.space, Global region is https://ai.inf.space
curl "https://cn.inf.space/v1/images/generations" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "doubao-seedream-5-0-pro",
    "prompt": "A white ceramic mug on a gray tabletop, soft natural light, e-commerce hero-image style",
    "size": "2K",
    "aspectRatio": "3:4"
  }'

Multi-image reference

Up to 10 reference images; refer to them in the prompt in upload order as "image 1 / image 2":

# China region, accelerated route; China international route is https://global.inf.space, Global region is https://ai.inf.space
curl "https://cn.inf.space/v1/images/edits" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY" \
  -F "model=doubao-seedream-5-0-pro" \
  -F "image[]=@bag.png" \
  -F "image[]=@scarf.png" \
  -F "prompt=Combine the handbag from image 1 and the scarf from image 2 in one product shot, soft studio lighting" \
  -F "size=2048x2048"

Reference images can also be passed through JSON images as public links, data: URLs, or base64; see Image generation and editing · Passing reference images.

Parameters

Parameter	Description
`model`	`doubao-seedream-5-0-pro`
`prompt`	Generation instruction; Chinese is supported natively
`size`	`1K` / `2K` tier, or a `widthxheight` from the table below
`aspectRatio`	Such as `"3:4"`, used together with a tier; effective for both text-to-image and editing
`image[]` / `images`	Reference images, up to 10
`n`	Number of images, `1`–`4`
`output_format`	Defaults to `jpeg`; options are `png` / `jpeg` / `webp`
`response_format`	`url` (default) or `b64_json`

quality, background, and mask have no effect on this model.

Sizes

Aspect ratio	1K	2K
1:1	`1024x1024`	`2048x2048`
4:3	`1152x864`	`2048x1536`
3:4	`864x1152`	`1536x2048`
16:9	`1424x800`	`2048x1152`
9:16	`800x1424`	`1152x2048`
3:2	`1248x832`	`2048x1376`
2:3	`832x1248`	`1376x2048`
21:9	`1792x768`	`2048x878`

Sizes outside the table are aligned by aspect ratio to the nearest specification; the top-level size in the response is the actual pixel size. This model does not support 4K.

Latency and recommendations

A single image takes roughly 60–120 seconds, longer with more reference images. Set synchronous timeouts to at least 180 seconds; the more robust approach is to use async jobs (the request body is exactly the same as the synchronous one; just submit it to POST /api/jobs).
For service-busy errors such as "An error occurred while processing your request.", tell the user to try again later instead of retrying repeatedly.
Copy the image promptly after receiving a url; the link is time-limited.

Pricing

Billed per image. Actual prices are shown in the console, with organization-specific prices taking precedence.