API
Embeddings
Create governed vector representations without a product data store
POST /v1/embeddings performs transient inference. Router routes the request and returns compatible vectors and usage; it does not persist, index or attach them to a Documents, Memory or Knowledge product.
Create embeddings
Use a route-scoped workspace credential and an embedding-capable model visible to that workspace.
curl https://api.octoryn.dev/v1/embeddings \
-H "Authorization: Bearer $OCTORYN_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"provider/embedding-model","input":["first text","second text"]}'Input and optional controls
Input may be one value or a batch supported by the selected route. dimensions, encoding_format and multimodal forms are accepted only where the eligible adapter and model declare support.
Discover embedding models
GET /v1/embeddings/models returns the tenant-visible embedding directory. Treat it as dynamic catalogue data and do not infer support from a marketing model name.
GET https://api.octoryn.dev/v1/embeddings/models
Authorization: Bearer $OCTORYN_API_KEYResponse and usage
The compatible response contains ordered embedding data plus normalized usage. Embeddings are non-streaming; apply explicit batch, input-size and application timeout limits.