Image generation

/

POST /v1/images/generations answers in one response rather than as a task. model and prompt are required; n, size and quality are optional and the models fall back to their own defaults. The reply is {"created": …, "data": [ … ]}, where each entry carries a url — b64_json is present but empty unless you ask for it with "response_format": "b64_json". Editing an existing image goes to POST /v1/images/edits with the source in image. Billing is per generated image, so n multiplies the charge.

POST /v1/images/generations 
curl https://api.chinaapi.ai/v1/images/generations \
  -H "Authorization: Bearer $CHINAAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "image-01",
    "prompt": "a red paper crane on a white desk, soft daylight",
    "n": 1
  }'

# {"created":1786620000,
#  "data":[{"url":"https://.../image_inference_output/....jpg","b64_json":""}],
#  "metadata":{"success_count":"1","failed_count":"0"}}
4K output

doubao-seedream-5-0-260128

With doubao-seedream-4-5-251128. Both reach 4K and both accept an input image for editing and multi-image composition, so size is worth setting explicitly rather than taking the default.

General purpose

wan2.7-image

With wan2.7-image-pro, which adds a 4K tier at roughly twice the price. Same request fields as the rest of this endpoint; the pro variant is the one to reach for when the output is going to be printed or upscaled.

Editing and low cost

step-image-edit-2

With step-2x-large and image-01. step-image-edit-2 is built for editing an image you supply and is the cheapest of the three; image-01 is the cheapest text-to-image model on the endpoint.

Tip This endpoint bills per call, not per token, and the price already covers one image — the usage log records a single prompt token, which is bookkeeping rather than a measure of your prompt. Asking for "n": 4 costs four times one image. Prices differ by more than an order of magnitude across the models here, so pick the model before tuning the prompt.