Embeddings
Text Embeddings
Convert one or more text inputs to vectors using an enabled embedding model. For a batch input, the response contains one vector per input.
Endpoint
POST
https://aiservice.prabisha.com/embeddings
Request
curl -X POST https://aiservice.prabisha.com/embeddings \
-H "X-API-Key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"input": ["First text to embed", "Another text to embed"],
"model": "gemini-embedding-2-preview",
"preferredProvider": "GEMINI"
}'
Fields
| Field | Type | Details |
|---|---|---|
| input | string or string[] | One text value or a batch of text values. Required. |
| model | string | Enabled embedding model ID. If omitted, the router selects an active default. |
| preferredProvider | string | Optional provider name, such as OPENAI or GEMINI. |
Response
{
"embeddings": [
[0.012, -0.034, 0.056],
[0.021, 0.018, -0.043]
],
"provider": "GEMINI",
"model": "gemini-embedding-2-preview",
"usage": {
"promptTokens": 0,
"totalTokens": 0
},
"providerId": "provider-id",
"providerModelId": "provider-model-id",
"latency": 240
}
Vector dimensions and token accounting depend on the selected provider and model.