Generate vector embeddings for text. Power RAG pipelines, semantic search, and document retrieval from Indian data centers.
POST https://api.forii.in/inference/v1/embeddings
Request
response = client.embeddings.create(
model="forii/embed-v3",
input=[
"What is the GST rate for textiles?",
"भारत में टेक्सटाइल का GST दर क्या है?"
]
)
print(len(response.data[0].embedding)) # 1024
print(response.usage) # {"prompt_tokens": 20, "total_tokens": 20}
const response = await client.embeddings.create({
model: "forii/embed-v3",
input: [
"What is the GST rate for textiles?",
"भारत में टेक्सटाइल का GST दर क्या है?",
],
});
console.log(response.data[0].embedding.length); // 1024
console.log(response.usage); // { prompt_tokens: 20, total_tokens: 20 }
curl https://api.forii.in/inference/v1/embeddings \
-H "Authorization: Bearer $FORII_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "forii/embed-v3",
"input": [
"What is the GST rate for textiles?",
"भारत में टेक्सटाइल का GST दर क्या है?"
]
}'
Response
{
"object": "list",
"data": [
{
"object": "embedding",
"index": 0,
"embedding": [0.0023, -0.0091, 0.0147, ...]
},
{
"object": "embedding",
"index": 1,
"embedding": [0.0041, -0.0078, 0.0082, ...]
}
],
"model": "forii/embed-v3",
"usage": {
"prompt_tokens": 20,
"total_tokens": 20
}
}
Parameters
| Parameter | Type | Required | Default | Description |
|---|
model | string | Yes | — | forii/embed-v3 |
input | string|array | Yes | — | Text or array of texts to embed |
dimensions | integer | No | 1024 | Output dimensions (variable-length embeddings) |
encoding_format | string | No | float | float or base64 |
Pricing
| Model | ₹/1K tokens |
|---|
forii/embed-v3 | ₹0.003 |
Embedding costs are extremely low. 1 million tokens costs ₹3. This makes bulk document indexing affordable for RAG pipelines.
Use cases
- RAG (Retrieval-Augmented Generation) — Index documents in Hindi and English, retrieve relevant context for chat completions
- Semantic search — Find similar documents across languages
- Clustering — Group similar content by embedding distance
- Deduplication — Detect near-duplicate documents
