Make it possible to set the only llama vision models for "Ask about screen content"

[giannisanni]

It would be handy if the model could be set to ask about screen content. this would differ from the default model, as I use another one for chat.