Generate an image

Create a synchronous image generation with POST /v1/images/generations. The returned image URL is copied to managed storage. Use a unique Idempotency-Key when retrying the same request.

curl https://api.mistrounter.ai/v1/images/generations \
  -H "Authorization: Bearer $MISTROUNTER_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: image-001" \
  -d '{
    "model": "qwen-image-3.0-pro",
    "prompt": "A premium product photograph on a misty lake",
    "size": "1024x1024"
  }'

For models whose inputs includes image, provide up to three public HTTPS references:

{
  "model": "qwen-image-3.0-pro",
  "prompt": "Keep the product and replace the background with a studio set",
  "image_urls": ["https://cdn.example.com/product.png"],
  "size": "2048x2048"
}

Supported common fields are model, prompt, image_urls, size, negative_prompt, prompt_extend, seed, and metadata. Read the model's inputs, sizes, availability, and pricing from GET /v1/models first.

Successful responses use the image.generation object and place durable URLs in output[].url. Failed submissions restore the reserved Tokens.