API Documentation

Integrate screengrasp into your applications

Access powerful click position prediction models through our simple API. Choose between:

Note: API access requires a Pro or Enterprise subscription plan.

Your Plan

Loading...
Change Plan

API Credits

Loading...

Your API Key

API Endpoints

Create Analysis Task

POST https://screengrasp.onrender.com/api/createAnalysisTask

Creates a new image analysis task and returns a task ID.

Important: Our servers automatically enter sleep mode during periods of inactivity. The first API call after such a period may take several minutes while the server boots up. Subsequent calls will be significantly faster as long as the server remains active.

Parameters

  • image - The screenshot file to analyze
  • taskDescription - Description of what to find/click in the image
  • mechanism (optional) - The model to use. Options:
    • screengrasp (default)
    • anthropic-computer-use
    • gpt4o-with-omniparser
    • molmo
    • gemini-flash
    • gpt4o

Get Task Status

GET https://screengrasp.onrender.com/api/getTaskStatus/:taskId

Retrieves the current status and results of an analysis task.

Response

  • status - Current task status:
    • QUEUED - Task is waiting in queue
    • COMPLETED - Analysis finished successfully
    • FAILED - Analysis failed
  • predictedClickPosition - When completed, contains the click coordinates
  • error - Error message if status is FAILED
Implementation Note: This endpoint should be polled regularly (every 500ms) until either:
  • The status is COMPLETED or FAILED
  • A timeout is reached (recommended: 5 minutes)
If you don't poll for >30 seconds, the task might expire and get deleted from the task queue. See the code examples below for implementation details in various languages.

Code Examples

Token Usage

Each API call consumes token credits from your account. Average token usage:

Token estimation calculator coming soon!