Embeddings

In one line

Send text and get back one vector per input, ready for a vector database, semantic search or clustering. The model is paranine/qwen3-embedding, on its own endpoint with the same key as every other model.

Request

POST https://sovereign.paranine.com/v1/embeddings

input is a string, or an array of strings to embed several at once.

curl https://sovereign.paranine.com/v1/embeddings \
  -H "Authorization: Bearer $P9_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "paranine/qwen3-embedding",
    "input": ["Paranine bills in rupees.", "Where is my order?"]
  }'

Response

{
  "object": "list",
  "data": [
    {"object": "embedding", "index": 0, "embedding": [0.0123, -0.0456, …]},
    {"object": "embedding", "index": 1, "embedding": [0.0789, 0.0012, …]}
  ],
  "model": "paranine/qwen3-embedding",
  "usage": {"prompt_tokens": 15, "total_tokens": 15}
}

data holds one entry per input, in the order you sent them. Each vector has 4,096 numbers.

Pricing

You're billed on usage.prompt_tokens.

Good to know

• Input length: each input can be up to 32,768 tokens.
• Request size: up to 2 MB per request; split larger batches.
• Rate limit: embeddings requests from all of your organisation's keys share one per-minute limit. Past it they answer 429 (rate_limited_org) with a Retry-After header.
• Residency: the model runs in the PK NPU pool, inside Pakistan.