Skip to main content
POST
Create embeddings

Authorizations

Authorization
string
header
required

The universal platform key, prefixed sk-conare-. The same bearer is provisioned by digest into ConareDB and accepted by the stateless embeddings endpoint, so database-backed search needs only one key.

Body

application/json
input
string[]
required

Texts to embed, each at most 8192 characters.

Required array length: 1 - 96 elements
Required string length: 1 - 8192
input_type
enum<string>
default:document

document creates passage-space vectors to store. query creates dated query-space vectors to search them. DR2E5 is asymmetric, so the two roles are not interchangeable.

Available options:
query,
document
model
string

Optional pin on the expected embedding model. The endpoint serves exactly one model; naming any other is a 400 (model_not_served), never a silent substitution.

Minimum string length: 1

Response

One vector per input, in input order.

success
boolean
required
model
string
required

Identifier of the serving embedding model.

dims
integer
required

Dimensionality of every returned vector.

Example:

1024

embeddings
number[][]
required

One unit-normalized vector per input, in input order.

usage
object
required