The Limitations of GPT for in Receiving Input Images as Input

The capability of GPT for in receiving input images as input this allows of course more dynamic interactions though it's not yet best of my knowledge available to the public so it's available only via an API for developers. There are some graphs that I would like to of course show is impossible I'm not GPT for I cannot make images out of a podcast but definitely what I wouldlike to discuss is some of them let's say benchmarks that have been reported on the technical report of OpenAI themselves which in my opinion is something that deserves our attention.