this strategy will use a screenshot of the page/locator with some processing to extract the needed data. should be used when the information you’re trying to extract is not present in the dom as a text but can be identified visually.

​

Properties

​

model

model: 
  | "claude-3-opus"
  | "claude-3-sonnet"
  | "claude-3.5-sonnet"
  | "claude-3-haiku"
  | "gpt4-turbo"
  | "gpt-4o";

the model to use in the extraction process.

​

type

type: "IMAGE";

the type of the strategy