Implement Multi-Modal Completions (Vision)

[Twister915]

Problem

I would like to build a tool that submits text and images to the OpenAI endpoints, so that I can implement some content moderation.

The Vision API is specified here.

I would like to implement this, as well, so I will use this issue to track my progress and reference it in my PRs. Not sure if this conflicts with any existing work, which is why I submit this issue.

Solution

Implement the Vision API

Alternatives

No response

[AlbertMarashi]

Go forward. Happy to review when ready