[Feature] Add multimodal support

[Barshan-Mandal]

trafficstars

Feature Request

Add video ,audio and image input for better multimodal q&a. Many pc cant run chatrtx due to some requirements.So these has become mandatory for our day to day life.Even chatrtx support describing YouTube video.

Its would be very good if you support both multimodal input and output for text ,video ,audio and image. They will work model specifically but you may add multimodal support by tricking multiple llm working for a prompt sequencially.