CLIP is the vision model used for indexing media: https://openai.com/index/clip/