Create Embeddings
Last updated
Last updated
input
string or array
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.
model
string
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.
encoding_format
string
The format to return the embeddings in. Can be either float
or base64
.
dimensions
integer
The number of dimensions the resulting output embeddings should have. Only supported in text-embedding-3
and later models.
user
string
NOT SUPPORTED