What model is better for text search embedding: text-search-davinci-query-001 or text-embedding-ada-002.
This link Introducing text and code embeddings
says that the text-search-davinci-query-001 is the best text search model but this model doesn’t appear anywhere else over the whole documentation.
Legacy models retiring in January have had their documentation nearly wiped clean off the site.
davinci requires 8x as much storage of vectors.
You can decide for yourself the next three months of model availability if you want to experiment with the departing davinci – at 2000 times the cost:
text-embedding-ada-002: $0.0001 / 1K tokens
text-similarity-davinci-001: $0.2000 / 1K tokens
I need model exactly for text search. The well known model text-embedding-ada-002 is advertised for all existing cases, like the cheapest and one-fit-all model.
Anyone tried embedding with other libraries?