Seedream 5.5 API Guide: Pricing, Parameters and Examples

What Seedream 5.5 does

Seedream 5.5 is an image model on HiAPI. One model ID, seedream-5.5, covers both text-to-image and image editing: send only a prompt and you get a new image; add reference images and the model edits or preserves what is in them. Both modes use the same endpoint and the same parameters. The only difference is whether the request includes image_urls.

Each call returns one image. This guide covers pricing, parameters, a working request and the current limits. All numbers reflect the API definition at the time of writing.

Seedream 5.5 API pricing

You pay a base price for the output size, plus a fee for each reference image.

• 1K / 1.5K: $0.0186 per image
• 2K (default): $0.0325 per image
• Reference images: $0.0075 each, including the first one, up to 10

Some worked examples:

• 1K text-to-image, no references: $0.0186
• 2K with one reference image: $0.0325 + $0.0075 = $0.04
• 2K with ten reference images (the maximum): $0.0325 + $0.075 = $0.1075

If you leave out size, the image is generated and billed at 2K. For drafts or small thumbnails, setting "size": "1K" explicitly cuts the base price by about 43%. Failed tasks are refunded in full.

Parameters

Everything except model and prompt is optional.

• prompt: required, 1–5,000 characters. Describe the image you want; when you pass references, say what to change and what to keep.
• size: 1K, 1.5K or 2K. Defaults to 2K.
• aspect_ratio: one of ten values: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 21:9, 2:1, 1:2. If omitted, the model picks one based on the prompt and references.
• image_urls: 1–10 publicly reachable http(s) image URLs. Omit it for pure text-to-image.
• output_format: png or jpeg. Defaults to png.
• background: auto, transparent or opaque. Defaults to auto. A transparent background requires png; combining it with jpeg is rejected.

Making a request

Image generation on HiAPI is asynchronous: you create a task, then poll for the result. model goes at the top level and everything else goes inside input.

Text-to-image at 2K, 16:9:

curl -X POST https://api.hiapi.ai/v1/tasks \
  -H "Authorization: Bearer $HIAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedream-5.5",
    "input": {
      "prompt": "A lighthouse on the coast at dawn, light fog, soft side lighting, photorealistic",
      "size": "2K",
      "aspect_ratio": "16:9"
    }
  }'

Editing a reference image at 1K with a transparent background:

{
  "model": "seedream-5.5",
  "input": {
    "prompt": "Keep the bottle and its label, remove the background, add soft top lighting",
    "image_urls": ["https://example.com/bottle.jpg"],
    "size": "1K",
    "background": "transparent",
    "output_format": "png"
  }
}

That second request costs $0.0186 + $0.0075 = $0.0261. Once the task is created, poll GET https://api.hiapi.ai/v1/tasks/{taskId} until it succeeds, then download the image URL from the output.

Current limits

• The maximum size is 2K. There is no 4K tier.
• Layered image output is not supported.
• Each call returns one image. For several images, send several tasks in parallel.
• Reference images must be public http(s) URLs. Local file paths and base64 are not accepted.
• Transparent backgrounds are PNG only.

Choosing a size

Use the default 2K for posters and product images that go straight to production. For internal reviews, style exploration in bulk, or small thumbnails, use 1K or 1.5K. They cost the same, so 1.5K is the better pick unless you specifically need 1K. Reference images are billed one by one, so pass only the ones that actually shape the result. For a product shot from several angles, start with one or two references and add more only if the output falls short.