Skip to main content
The /v1/embeddings endpoint converts text into numerical vectors that capture semantic meaning. You can embed a single string or a batch of strings in one request. The resulting vectors are suitable for semantic search, retrieval-augmented generation (RAG), clustering, and classification tasks.

​
Authentication

Include your API key in every request: 
Authorization: Bearer YOUR_API_KEY

​
Send an embeddings request

Provide an input (a string or array of strings) and a model. TokModel routes the request to the specified embedding model and returns one vector per input item. 
curl https://tokmodel.com/v1/embeddings \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/text-embedding-3-small",
    "input": "The quick brown fox jumps over the lazy dog."
  }'

​
Embed multiple inputs in one request

Pass an array of strings to input to embed a batch in a single API call. The response contains one entry per input, in the same order. 
curl https://tokmodel.com/v1/embeddings \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/text-embedding-3-small",
    "input": [
      "How do I reset my password?",
      "Where can I find my invoices?",
      "How do I cancel my subscription?"
    ]
  }'

​
Example response

Each object in the data array corresponds to one input string. The embedding field contains the raw float vector. 
{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [
        0.0023064255,
        -0.009327292,
        0.015797347,
        "..."
      ]
    }
  ],
  "model": "openai/text-embedding-3-small",
  "usage": {
    "prompt_tokens": 10,
    "total_tokens": 10
  }
}
The embedding array above is truncated for readability. Real vectors typically contain 512 to 3072 dimensions depending on the model.

​
Compute cosine similarity

After embedding two pieces of text, compare them with cosine similarity. A score close to 1.0 means the texts are semantically similar.
python
import numpy as np

def cosine_similarity(a, b):
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

response = client.embeddings.create(
    model="openai/text-embedding-3-small",
    input=[
        "How do I reset my password?",
        "I forgot my password, what should I do?",
    ],
)

vec_a = response.data[0].embedding
vec_b = response.data[1].embedding

score = cosine_similarity(vec_a, vec_b)
print(f"Similarity: {score:.4f}")  # e.g. 0.9312

​
Common use cases

Semantic search — embed a user query and compare it against pre-embedded documents to find the most relevant results, even when the exact words differ. Retrieval-augmented generation (RAG) — embed your knowledge base and retrieve the top-k matching chunks before passing them to a chat model as context. Clustering — group similar documents together by clustering their vectors using algorithms like k-means without any labeled training data. Classification — train a lightweight classifier on top of embeddings to categorize text into predefined labels.