Support for Vision

Thanks for the project. It's delightfully simple and a pleasure to use. Currently, Messages take a role and a string for the content. When it comes to vision though I believe we need to pass more than a string in.

Is there any plans to support this or if its possible already?