Docs / API reference

Embeddings

The embeddings endpoint lives at /v1/embeddings and is equally compatible with the official SDK.

PYTHON

v = client.embeddings.create(
    model="text-embed-3",
    input=["first passage", "second passage"],
)

Model

Dimensions & price

text-embed-3

1536 dims · $0.02 / 1M tokens

bge-m3

1024 dims · $0.008 / 1M tokens

qwen-embed-2

1024 dims · $0.006 / 1M tokens

For bulk ingestion, keep each request under 256 inputs; larger batches are split into multiple upstream requests billed separately.