Seedream (Doubao) family

ByteDance’s Doubao image models, three in total, called through the OpenAI-compatible endpoint /v1/images/generations; both text to image and editing use this endpoint.

Choosing a model

For live prices, see the model page .

Text to image

import base64
from openai import OpenAI
 
client = OpenAI(api_key="YOUR_OFOX_API_KEY", base_url="https://api.ofox.ai/v1", timeout=600)
 
resp = client.images.generate(
    model="volcengine/doubao-seedream-5.0-pro",
    prompt="A red apple on a white table",
    size="2560x1440",
)
 
with open("output.jpg", "wb") as f:
    f.write(base64.b64decode(resp.data[0].b64_json))

curl https://api.ofox.ai/v1/images/generations \
  --max-time 600 \
  -H "Authorization: Bearer $OFOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "volcengine/doubao-seedream-5.0-pro",
    "prompt": "A red apple on a white table",
    "size": "2560x1440"
  }' | jq -r '.data[0].b64_json' | base64 --decode > output.jpg

The output format is always JPEG, so save files with a .jpg extension. Base64 is returned by default; pass "response_format": "url" to get an image link instead.

Size (size)

size accepts two forms:

• Resolution presets 1k, 2k, 4k (case-insensitive): these set only the resolution, and the model chooses the aspect ratio based on the prompt. In our tests, 2k on 4.5 produced both 2048×2048 and 1664×2496.
• Exact dimensions WIDTHxHEIGHT: a fixed aspect ratio, which must fall within the pixel range in the table below.

”—” means not measured in this round; refer to the Volcano Engine Ark official docs .

Commonly used valid sizes (meeting the minimum for 4.5 and 5.0 Lite):

Image editing & multiple reference images

import base64
from openai import OpenAI
 
client = OpenAI(api_key="YOUR_OFOX_API_KEY", base_url="https://api.ofox.ai/v1", timeout=600)
 
with open("apple.jpg", "rb") as f:
    ref = "data:image/jpeg;base64," + base64.b64encode(f.read()).decode()
 
resp = client.images.generate(
    model="volcengine/doubao-seedream-5.0-pro",
    prompt="Make the apple purple, keep everything else the same",
    size="2k",
    extra_body={"image": [ref]},
)
 
with open("apple_purple.jpg", "wb") as f:
    f.write(base64.b64decode(resp.data[0].b64_json))

# The base64 is long; putting it straight on the command line fails with
# "Argument list too long", so write it to a file first
printf 'data:image/jpeg;base64,%s' "$(base64 < apple.jpg | tr -d '\n')" > apple.b64
jq -n --rawfile img apple.b64 '{
  model: "volcengine/doubao-seedream-5.0-pro",
  prompt: "Make the apple purple, keep everything else the same",
  size: "2k",
  image: [$img]
}' > body.json
 
curl https://api.ofox.ai/v1/images/generations \
  --max-time 600 \
  -H "Authorization: Bearer $OFOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d @body.json | jq -r '.data[0].b64_json' | base64 --decode > apple_purple.jpg

• image is an array; each item is an image URL or a string starting with data:image/...;base64,. Put multiple reference images into the array in order, and describe the role of each image in the prompt.
• input_images is a Qwen-Image-only field; Seedream does not support it.
• Seedream does not support /v1/images/edits; to edit images, use the image field described in this section.

Image sets & multiple images

Image set example:

{
  "model": "volcengine/doubao-seedream-5.0-lite",
  "prompt": "Generate 3 separate images: the same fox in spring, summer and autumn",
  "size": "2k",
  "sequential_image_generation": "auto",
  "sequential_image_generation_options": { "max_images": 3 }
}

• auto means the model decides how many images to generate; max_images is only an upper limit. In our tests, with the same request 4.5 combined a three-panel comic into a single image and returned just 1. If you need a fixed number, state “generate N separate images” explicitly in the prompt.
• Each image is billed separately; usage.output_tokens is the total across all images.

Watermarks

In our tests, calls through OfoxAI have no watermark by default; no extra parameter is needed.

Latency

We measured about 12 to 40 seconds at 2K, and about 76 seconds for a set of 3 images. Set the client timeout to 300 seconds or more.

Common errors

The Request id: 0217… at the end of an error message is the upstream request ID; when contacting support, include it together with the request_id from the response.

The parameter `size` specified in the request is not valid: image size must be at least 3686400 pixels.
The parameter `size` specified in the request are not valid: image area must be at least 921600 pixels.
The parameter `size` specified in the request is not valid: image size must be at most 16777216 pixels.
The parameter `size` specified in the request is not valid: the specified size is not supported for model doubao-seedream-4-5.
The parameter `size` specified in the request are not valid: must be 'WIDTHxHEIGHT' or a supported size preset.
The parameter `size` specified in the request is not valid: size must be one of 'WIDTHxHEIGHT', '1k', '2k', or '4k'.
The parameter `sequential_image_generation` specified in the request are not valid: the parameter `sequential_image_generation` is not supported by the current model.
The parameter `output_format` specified in the request is not valid: the parameter `output_format` is not supported by the current model.

Official docs

The official docs describe behavior when calling Volcano Engine Ark directly; when calling through OfoxAI, the measured results on this page take precedence.

• Volcano Engine Ark documentation 

FAQ