A semantic cache is just an encoder in front of a regular cache. It is not hard to connect the two, so where do you see the value? If you DIY, you can choose your own cache.

Isn't this how most vector search backends work, cosine similarity sorting against whatever the vector embeddings returned? How is this different?