ID of the model to use. You can use the List models API to see all of your available models, or see our Model overview for descriptions of them.
Optional
dimensionsThe number of dimensions the resulting output embeddings should have. Only
supported in text-embedding-3
and later models.
Optional
encoding_The format to return the embeddings in. Can be either float
or
base64
.
Optional
userA unique identifier representing your end-user, which can help OpenAI to monitor and detect abuse. Learn more.
Input text to embed, encoded as a string or array of tokens. To embed multiple inputs in a single request, pass an array of strings or array of token arrays. The input must not exceed the max input tokens for the model (8192 tokens for
text-embedding-ada-002
), cannot be an empty string, and any array must be 2048 dimensions or less. Example Python code for counting tokens.