Embeddings

POST /v1/embeddings

OpenAI-compatible request and response shape.

Request

{
  "model": "text-embedding-3-small",
  "input": "The text to embed"
}

Accepts the same headers as Chat Completions — x-op-team and x-op-policy apply here too, and requests pass through the same pipeline (auth, rate limit, policy, cache, route, log).

Response

Standard OpenAI-shaped embedding vector response, plus x-openproxyai-* gateway headers — see API Reference Overview.

Caching

Embedding requests are cache-eligible at the L1/L2 exact-match tiers. They are not matched by the L3 semantic tier, since semantic matching itself depends on embeddings — see Caching.