Embeddings

BYOK embeddings with 30-day Redis cache. OpenAI, Cohere, Voyage AI.

POST /v2/embed accepts a single string or an array of strings and returns a list of embedding vectors. Results are cached in Redis for 30 days.

Providers

OpenAI - text-embedding-3-small T1 ($0.02/1M) · text-embedding-ada-002 T2 ($0.10/1M) · text-embedding-3-large T3 ($0.13/1M)
Cohere - embed-multilingual-v3.0 T2 ($0.10/1M)
Voyage AI - voyage-3-lite T1 ($0.02/1M) · voyage-3 T2 ($0.06/1M) · voyage-large-2 T3 ($0.12/1M)

Cache

Cache key: SHA-256 of {sorted(inputs), provider, model, encoding_format}. Cached for 30 days. The cached: true field in the response tells you if the result came from cache (no provider call made, zero cost).

{
  "input": "text or array of texts",
  "provider": "openai | cohere | voyage",
  "model": "text-embedding-3-small",
  "encoding_format": "float | base64",
  "dimensions": 1536
}