Animate a product photo with Agent Completions: $0.625 on Wan 3.0

An Agent Completion can take one product photo as an input_image part and animate it into a 5-second Wan 3.0 clip at 720p for $0.625 ($0.125 per second). A generation_spend_cap_usd of 1 covers it, and images are the only attachment type the endpoint accepts.

Clip price by resolution

Wan 3.0 accepts 2 to 30 seconds.

The body

Use messages with an input_image part. The image URL must be HTTPS or an asset_id, never both. An image can be up to 30 MB.

{
  "messages": [
    { "role": "user", "content": [
      { "type": "input_text", "text": "Animate this mug photo into a 5 second Wan 3.0 clip at 720p." },
      { "type": "input_image", "image_url": "https://cdn.example.com/mug.jpg" }
    ] }
  ],
  "generation_spend_cap_usd": 1
}

Limits and gotchas

Attachments allow up to 30 images, 500 MB in total; PDFs and other files are not accepted yet.

• A bad URL or hotlink-protected host gives 502 attachment_fetch_failed.
• Assistant turns in messages are rejected with 400.
• Each completion runs in a new thread; there is no conversation continuation.