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

# POST /v1/embeddings

> OpenAI-compatible text embeddings for retrieval, clustering, classification and RAG.

OpenAI-compatible endpoint to generate text embeddings — dense vectors of a text for retrieval, clustering, classification or RAG. Billed per input token.

## Available models

| Model                           | Dimensions | Price /1M tokens |
| ------------------------------- | ---------- | ---------------- |
| `openai/text-embedding-3-small` | 1536       | \$0.02 USD       |
| `openai/text-embedding-3-large` | 3072       | \$0.13 USD       |

## Basic example

```bash theme={null}
curl -X POST https://api.geekhub.mx/v1/embeddings \
  -H "Authorization: Bearer ghub_sk_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/text-embedding-3-small",
    "input": "The quick brown fox jumps over the lazy dog"
  }'
```

Response:

```json theme={null}
{
  "object": "list",
  "model": "openai/text-embedding-3-small",
  "data": [
    { "object": "embedding", "index": 0, "embedding": [0.012, -0.043, ...] }
  ],
  "usage": {
    "prompt_tokens": 12,
    "total_tokens": 12,
    "cost_usd": 0.00000025,
    "cost_mxn": 0.000005
  }
}
```

## Batching

`input` also accepts an array (up to 2048 items):

```json theme={null}
{
  "model": "openai/text-embedding-3-large",
  "input": [
    "First document…",
    "Second document…",
    "Third…"
  ]
}
```

`data` in the response is ordered to match the input.

## Reduce dimensions

`-3-*` models let you request a lower dimensionality (useful to shrink storage in your vector DB):

```json theme={null}
{
  "model": "openai/text-embedding-3-large",
  "input": "Text…",
  "dimensions": 512
}
```

**Note:** price is per input token, not per dimension. Reducing dimensions doesn't reduce the embedding cost — only your storage cost and search latency.

## Zero Data Retention

Same rules as `/chat/completions` — set `zdr: true` or configure your guardrail. OpenAI's text-embedding-3-\* is on the verified list.

## Managed vector store

Coming: `POST /v1/vector-stores` for hosted collections, upload and semantic query without running your own Pinecone/pgvector. For now, use the endpoint directly and store vectors wherever your infrastructure lives.