Embeddings¶

POST /v1/embeddings - converts text into vectors for search, clustering and retrieval.

Request¶

curl https://<endpoint>/v1/embeddings \
  -H "Authorization: Bearer <key>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<embedding-model-id>",
    "input": "The quick brown fox."
  }'

Python¶

response = client.embeddings.create(
    model="<embedding-model-id>",
    input=["first document", "second document"],
)
vectors = [d.embedding for d in response.data]

input accepts a string or an array. Batching is significantly more efficient than one request per document.

Notes¶

Embedding models are served on a separate internal path from generative models and do not pass through the inference gateway. This is transparent to clients.
Vector dimensions are model-specific. Do not mix vectors from different models in one index.
Embedding model IDs appear in GET /v1/models alongside generative ones. See Model catalog.