Skip to main content
Wanxiang Image Generation & Editing 2.7 (Wan2.7 Image) is Alibaba’s latest multimodal image model, available through Starrise AI. Supports text-to-image, image editing, interactive region editing, and multi-image generation with up to 4K output resolution.

​
Model List

Model IDDescription
wan2.7-image-proProfessional version, text-to-image supports 4K HD output
wan2.7-imageStandard version, faster generation speed

​
Key Capabilities

  • Text-to-image — Generate images from text descriptions
  • Image editing — Edit existing images with text prompts (supports 0–9 input images)
  • Interactive editing — Specify a bounding box region via bbox_list for localized editing
  • Multi-image generation — Generate multiple related images in a single request (enable_sequential)
  • Thinking mode — Enhanced reasoning to improve text-to-image quality

​
Quick Example

curl https://ai.alad.com/v1/images/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "wan2.7-image-pro",
    "prompt": "A flower shop with elegant windows, beautiful wooden doors, and flowers on display",
    "size": "2K",
    "thinking_mode": true
  }'

​
Parameters

ParameterTypeRequiredDescription
modelstringYeswan2.7-image-pro or wan2.7-image
promptstringYesImage description text, up to 5000 characters
imagestringNoInput image URL or Base64 data URI for editing. Supports JPEG, JPG, PNG, BMP, WEBP, max 20 MB, up to 9 images
nintegerNoNumber of images to generate. Sequential mode off: 1–4, default 1; sequential mode on: 1–12, default 12
sizestringNo1K, 2K (default), or 4K (wan2.7-image-pro text-to-image only). Also supports custom {width}x{height} pixel values
enable_sequentialbooleanNoEnable multi-image output mode, default false
thinking_modebooleanNoEnable thinking mode to improve quality, default true. Only effective when there are no image inputs and sequential mode is off
watermarkbooleanNoWhether to add a watermark, default false
seedintegerNoRandom seed [0, 2147483647], same seed produces similar results
response_formatstringNourl (default) or b64_json

​
Resolution Reference

SpecPixels
1K1024×1024
2K2048×2048
4K4096×4096 (wan2.7-image-pro text-to-image only)
When image input is provided, the output aspect ratio matches the input image and is scaled to the selected resolution.

​
Response Fields

FieldDescription
data[].urlGenerated image URL (PNG, valid 24 hours)
data[].b64_jsonBase64-encoded image (only returned when response_format=b64_json)
data[].revised_promptActual prompt after AI expansion
createdCreation timestamp

API Reference

View the interactive API Playground for Wan2.7 Image.