> ## Documentation Index
> Fetch the complete documentation index at: https://docs.inferoute.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# POST /v1/embeddings

> Generate vector embeddings for text input. Accepts input text and model, returns embedding vectors for semantic search, RAG pipelines, and similarity tasks.

The embeddings endpoint converts text into high-dimensional numeric vectors that capture semantic meaning. You can use these vectors to build retrieval-augmented generation (RAG) pipelines, power semantic search, cluster documents by topic, or classify text without fine-tuning.

## Endpoint

```
POST https://api.inferoute.ai/v1/embeddings
```

## Request parameters

<ParamField body="model" type="string" required>
  The embedding model to use. For example: `text-embedding-3-small` or `text-embedding-3-large`. Retrieve the full list of available embedding models from [GET /v1/models](/docs/api-reference/models).
</ParamField>

<ParamField body="input" type="string | string[]" required>
  The text to embed. Pass a single string for one embedding or an array of strings to embed multiple inputs in a single request. Arrays are more efficient than making one request per string.
</ParamField>

<ParamField body="encoding_format" type="string" default="float">
  Format for the returned vectors. Use `"float"` for a JSON array of numbers (default), or `"base64"` for a base64-encoded binary representation that reduces response size.
</ParamField>

<ParamField body="dimensions" type="integer">
  Number of dimensions in the output embedding. Only supported by models that accept a dimensions parameter (e.g., `text-embedding-3-small`, `text-embedding-3-large`). Reducing dimensions trades some accuracy for lower storage and compute cost.
</ParamField>

## Response fields

<ResponseField name="object" type="string">
  Always `"list"`.
</ResponseField>

<ResponseField name="data" type="object[]">
  <Expandable title="properties">
    <ResponseField name="index" type="integer">
      Zero-based index corresponding to the input at that position.
    </ResponseField>

    <ResponseField name="object" type="string">
      Always `"embedding"`.
    </ResponseField>

    <ResponseField name="embedding" type="number[]">
      The embedding vector as an array of floating-point numbers. The length equals the model's output dimension (or the value of `dimensions` if specified).
    </ResponseField>
  </Expandable>
</ResponseField>

<ResponseField name="model" type="string">
  The model that generated the embeddings.
</ResponseField>

<ResponseField name="usage" type="object">
  <Expandable title="properties">
    <ResponseField name="prompt_tokens" type="integer">
      Total tokens across all inputs.
    </ResponseField>

    <ResponseField name="total_tokens" type="integer">
      Same as `prompt_tokens` for embedding requests (no completion tokens are generated).
    </ResponseField>
  </Expandable>
</ResponseField>

## Examples

### Single string input

<CodeGroup>
  ```bash curl theme={null}
  curl https://api.inferoute.ai/v1/embeddings \
    --request POST \
    --header "Authorization: Bearer YOUR_API_KEY" \
    --header "Content-Type: application/json" \
    --data '{
      "model": "text-embedding-3-small",
      "input": "Inferoute routes your LLM requests intelligently."
    }'
  ```

  ```python Python theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.inferoute.ai/v1",
  )

  response = client.embeddings.create(
      model="text-embedding-3-small",
      input="Inferoute routes your LLM requests intelligently.",
  )
  vector = response.data[0].embedding
  print(f"Embedding dimension: {len(vector)}")
  ```
</CodeGroup>

**Response:**

```json theme={null}
{
  "object": "list",
  "data": [
    {
      "index": 0,
      "object": "embedding",
      "embedding": [0.0023064255, -0.009327292, 0.015797043, "..."]
    }
  ],
  "model": "text-embedding-3-small",
  "usage": {
    "prompt_tokens": 9,
    "total_tokens": 9
  }
}
```

### Array input (batch embedding)

<CodeGroup>
  ```bash curl theme={null}
  curl https://api.inferoute.ai/v1/embeddings \
    --request POST \
    --header "Authorization: Bearer YOUR_API_KEY" \
    --header "Content-Type: application/json" \
    --data '{
      "model": "text-embedding-3-small",
      "input": [
        "How do I reset my password?",
        "Where can I find my invoice?",
        "How do I upgrade my plan?"
      ]
    }'
  ```

  ```python Python theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.inferoute.ai/v1",
  )

  texts = [
      "How do I reset my password?",
      "Where can I find my invoice?",
      "How do I upgrade my plan?",
  ]

  response = client.embeddings.create(
      model="text-embedding-3-small",
      input=texts,
  )

  for item in response.data:
      print(f"Index {item.index}: {len(item.embedding)} dimensions")
  ```
</CodeGroup>

**Response:**

```json theme={null}
{
  "object": "list",
  "data": [
    {"index": 0, "object": "embedding", "embedding": [0.0023, -0.0093, "..."]},
    {"index": 1, "object": "embedding", "embedding": [-0.0041, 0.0187, "..."]},
    {"index": 2, "object": "embedding", "embedding": [0.0112, -0.0034, "..."]}
  ],
  "model": "text-embedding-3-small",
  "usage": {
    "prompt_tokens": 27,
    "total_tokens": 27
  }
}
```

## Common use cases

* **RAG pipelines** — embed your documents at index time, embed user queries at runtime, and retrieve the closest chunks by cosine similarity before passing them to a chat completions request.
* **Semantic search** — find documents that are conceptually similar to a query even when no keywords match.
* **Document clustering** — group large collections of text by topic without labeled training data.
* **Classification** — train a lightweight classifier on top of embeddings rather than fine-tuning a full model.

<Tip>
  Batch your inputs into a single request whenever possible. Sending 100 strings in one array call is significantly faster and cheaper than making 100 individual requests.
</Tip>