Embeddings

POST /v1/embeddings — same shape as OpenAI's /v1/embeddings. Pass an input (string or array of strings) and a model id; the response returns one embedding vector per input.

curl -sS https://api.alt.advania.ai/v1/embeddings \
  -H "Authorization: Bearer $ALT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "advania-embed-small",
    "input": ["hello world"]
  }'

The embedding model used for a Knowledge Base must match the model used for indexing that KB — mixed dimensionality is not supported.

Full request/response shape in the interactive OpenAPI.