oup_oup_andaway, we chose not to display that information because it does not impact the user's experience, and to keep the interface concise and easy to understand.
If the model supports image inputs (or any other modalities), we will pass the image directly to it. If not, we will use another model that can process images, then send the parsed output to the chosen model. For the user, the result remains the same regardless of the method used.
xigoi
It seems that the cheatsheet is not available on mobile. I think it should be.
Ideally it would also be displayed on mobile, but since there isn't an obvious way to implement it, we felt it was a good compromise to keep it on desktop only.
