Embeddings

Use embeddings to convert text into vectors for search, retrieval, clustering, ranking, and similarity.

POST https://api.rout.my/v1/embeddings

Request

curl https://api.rout.my/v1/embeddings \
  -H "Authorization: Bearer $ROUTMY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "provider/embedding-model-id",
    "input": "The food was delicious and the service was excellent."
  }'

Request fields

Field	Type	Required	Notes
model	string	Yes	Exact embedding model ID from /v1/models.
input	string, array of strings, token array, or batch token arrays	Yes	Text or token input to embed.
encoding_format	string	No	float or base64, when supported by the upstream model.
dimensions	integer	No	Requested output dimension count, when supported by the model.
user	string	No	Client-side user identifier.

Extra fields are preserved for providers that support them.

Batch input

{
  "model": "provider/embedding-model-id",
  "input": [
    "First document.",
    "Second document.",
    "Third document."
  ]
}

The response preserves input order through the index field.

Python

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://api.rout.my/v1",
    api_key=os.environ["ROUTMY_API_KEY"],
)

response = client.embeddings.create(
    model="provider/embedding-model-id",
    input=["First document.", "Second document."],
)

for item in response.data:
    print(item.index, len(item.embedding))

Response

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [0.0023, -0.0093, 0.0142]
    }
  ],
  "model": "provider/embedding-model-id",
  "usage": {
    "prompt_tokens": 5,
    "total_tokens": 5
  }
}

Quota accounting

Embedding requests count prompt tokens only. The final quota usage is:

input tokens x token_multiplier

Use /v1/models to inspect each model's token_multiplier.