Dumpling AI Extract Image

55 credits/call
ae_dumplingai_extract_image

Extract text and insights from images using OCR and AI analysis via DumplingAI.

Image
55 credits/call

Pricing

55 credits/call
ConditionCharge
Each call55 credits/call

Parameters

Optional parameters use the platform defaults when omitted.

ParameterTypeRequiredDescriptionDefault
imagesarrayYesArray of image URLs or base64-encoded strings to analyze.—
promptstringYesInstructions that describe the insights to extract from the images.—
jsonModebooleanNoWhen true, requests the model to respond with JSON-formatted output.false
inputMethodenum (url | base64)YesIndicates whether binary content is supplied via URL or base64-encoded string.—
requestSourceenum (API | WEB | MAKE_DOT_COM | ZAPIER | N8N | PLAYGROUND | DEFAULT_AUTOMATION | AGENT_PREVIEW | AGENT_LIVE | AUTOPILOT | STUDIO)NoOptional identifier describing where the API request originated.—

Invocation example

Copy the request and explore its response example.

POST/agent-api/v1/tool/executeX-Api-Key
Request coderequest.sh
curl --request POST \
  --url 'https://agentearth.ai/agent-api/v1/tool/execute?tool_name=ae_dumplingai_extract_image' \
  --header "X-Api-Key: $AGENTEARTH_API_KEY" \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "params": {
    "images": [
      "https://file.gw.xnetwk.net/file/d9f65510-49e3-45fe-83dd-c72631de40e9.jpg",
      "https://file.gw.xnetwk.net/file/b8f3712d-d273-4f2a-a24f-f976b823e202.jpg",
      "https://file.gw.xnetwk.net/file/9944e74f-ae0f-4bf8-8ad4-c18f6f0a4ced.jpg",
      "https://file.gw.xnetwk.net/file/3b062f78-1044-488d-a692-1a143b0662d2.jpg",
      "https://file.gw.xnetwk.net/file/bbfff81a-5645-44bf-9d8e-22985f6712f8.jpg"
    ],
    "inputMethod": "url",
    "jsonMode": true,
    "prompt": "These are 5 numbered e-commerce product detail images for the XF45V retro game console (image 1 = ambient light, image 2 = vibration motor, image 3 = battery, image 4 = joystick, image 5 = grip). For EACH image, transcribe ALL visible text exactly as written, in reading order (top to bottom, left to right). Include every text element: large stylized headings, feature titles, small subtitle text, spec text, on-screen UI text, button labels, and any English or numeric text. Do not skip small or low-contrast text. Preserve original characters exactly (Chinese as Chinese, English as English). Output strictly as JSON: [{\"image\":1,\"lines\":[\"...\",...]},{\"image\":2,\"lines\":[\"...\"]},...]"
  }
}'
Response exampleresponse.json
JSONPartial content
{
  "creditUsage": 150,
  "imageCount": 5,
  "prompt": "These are 5 numbered e-commerce product detail images for the XF45V retro game console (image 1 = ambient light, image 2 = vibration motor, image 3 = battery, image 4 = joystick, image 5 = grip). For EACH image, transcribe ALL visible text exactly as writt",
  "results": [
    {
      "image": 1,
      "lines": [
        "360°"
      ]
    }
  ]
}

This example shows part of the response. Copied content is valid JSON.

Guides using this tool

See how this tool fits into real workflows