[Technotroll]

Does the vision-language-model process raw image data, or does it process OCR character output?

[bigfudge]

Gpt4v seems to be doing the former, at least in my experiments with it. It interprets plots and categorises images.