Embeddings · Gemini Native Format

Naming note

Some documentation classifies this endpoint as "Gemini native format." Based on its actual path and parameters, it is really OpenAI's earlier "engines" path style (the model ID goes in the path rather than the request body) — functionally identical to OpenAI native format's /v1/embeddings, just a different path shape.

Create embeddings with a specific engine

POST /v1/engines/{model}/embeddings

Path parameters

Parameter	Type	Required	Description
model	string	Yes	Model / engine ID, e.g. text-embedding-ada-002

Request parameters

Parameter	Type	Required	Description
input	string | array	Yes	Text to embed, or an array of texts (batch)
encoding_format	string	No	float (default) or base64
dimensions	integer	No	Output vector dimensions

Request example

curl https://ai.youqi.tech/v1/engines/text-embedding-ada-002/embeddings \
  -H "Authorization: Bearer sk-YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": "YouQi AI is a unified AI model gateway."
  }'

Response example

{
  "object": "list",
  "data": [
    { "object": "embedding", "index": 0, "embedding": [0.0023, -0.0091, 0.0157] }
  ],
  "model": "text-embedding-ada-002",
  "usage": { "prompt_tokens": 18, "total_tokens": 18 }
}

TIP

New projects should prefer OpenAI native format's /v1/embeddings. This endpoint mainly serves clients that depend on the legacy "engines" path.