OpenAI's smaller text-embedding-3 model, which OpenAI calls an improved, more performant version of its ada embedding model. It returns 1536-dimension vectors by default and costs less than its sibling text-embedding-3-large.
Pick it if
you want the hosted embedding model every Octomind plan includes, Free too
you want embeddings without loading model weights into your machine's memory
you want shorter vectors via the dimensions parameter to shrink your index
Skip it if
a single input runs past 8,192 tokens, the per-input limit
text-embedding-3-large is the larger of OpenAI's two third-generation text embedding models, returning 3072-dimension vectors by default. It sits above text-embedding-3-small, which costs less per token and returns 1536-dimension vectors.
Pick it if
retrieval quality matters more than cost per token
you want shorter vectors via the dimensions parameter, for example 256
you embed non-English as well as English text
Skip it if
text-embedding-3-small is accurate enough at a lower price
you need to embed images; it accepts text only
FAQ
Which is cheaper, OpenAI Embedding 3 Small or OpenAI Embedding 3 Large?
OpenAI Embedding 3 Small is cheaper: $0.02/M against $0.13/M.
Which handles longer documents?
Both take up to 8k tokens per text.
Can I switch from OpenAI Embedding 3 Small to OpenAI Embedding 3 Large without re-indexing?
No. Vectors from different embedding models are not comparable, so a switch means re-embedding the whole corpus.
Try either model
Switch between models inside the same Octomind session.