Image description query implementation

[ljaniszewski00]

It would be nice to have an option to make a request with gpt-4-vision-preview model in which we provide image and string prompt and it outputs a string answer to the prompt based on the given image. Please, correct me if this function already exists :)