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

# Introduction

> One endpoint per modality, every provider behind it.

The LLM Stats gateway is a thin, OpenAI-compatible router. You send a request to a single endpoint, we pick the best healthy provider for that model, and we transparently retry on failure — without changing your client code.

## Base URL

```
https://gateway.llm-stats.com
```

All endpoints documented in this section live under `/v1`.

## Authentication

Send your API key as a Bearer token on every request.

```bash theme={"theme":{"light":"github-light","dark":"github-dark"}}
Authorization: Bearer YOUR_API_KEY
```

Create and manage keys in the [LLM Stats dashboard](https://llm-stats.com).

<Tip>
  Keys starting with `ze_` are LLM Stats keys. Don't proxy upstream provider
  keys — you don't need them. Routing, retries, and accounting all happen on our
  side.
</Tip>

## How routing works

<Steps>
  <Step title="You send one request">
    A normal request to a modality endpoint with a `model` field — no provider
    prefix, no provider-specific quirks.
  </Step>

  <Step title="We pick a healthy provider">
    The router scores every provider serving that model on live latency,
    throughput, error rate, and capacity, then dispatches to the best one.
  </Step>

  <Step title="We fail over transparently">
    If the chosen provider degrades or errors mid-flight, we retry on the next
    best option. Your code only sees the final, successful response (or a clean
    error envelope if every provider fails).
  </Step>
</Steps>

## Discovering gateway-routable models

Use the Stats API to enumerate which models you can call through the gateway.
Every model returned by `GET /stats/v1/models` carries an `inference` block:

```json theme={"theme":{"light":"github-light","dark":"github-dark"}}
{
  "id": "gpt-5-2025-08-07",
  "inference": {
    "available": true,
    "endpoint": "https://gateway.llm-stats.com/v1/chat/completions",
    "gateway_model_id": "gpt-5-2025-08-07",
    "openai_compatible": true,
    "supports_streaming": true,
    "supports_tools": true,
    "supports_vision": false
  }
}
```

Pass `inference.gateway_model_id` as the `model` field in your gateway request.
See the [Stats API introduction](/api-reference/introduction#discovering-gateway-routable-models) for the field reference.

## Endpoints at a glance

| Modality       | Endpoint                    | Shape                                    |
| -------------- | --------------------------- | ---------------------------------------- |
| Chat / LLM     | `POST /v1/chat/completions` | OpenAI-compatible, sync or streaming     |
| Image / Video  | `POST /v1/generations`      | Unified async resource with long-polling |
| Text-to-speech | `POST /v1/tts/synthesize`   | Returns audio bytes                      |
| Speech-to-text | `POST /v1/stt/transcribe`   | Multipart upload, JSON transcript        |

<CardGroup cols={2}>
  <Card title="Chat completions" icon="message" href="/gateway/chat-completions">
    OpenAI-compatible LLM inference with streaming.
  </Card>

  <Card title="Image & video generations" icon="image" href="/gateway/generations">
    Single async API for image and video models.
  </Card>

  <Card title="Text-to-speech" icon="waveform" href="/gateway/text-to-speech">
    Synthesize audio from text.
  </Card>

  <Card title="Speech-to-text" icon="microphone" href="/gateway/speech-to-text">
    Transcribe audio files to text.
  </Card>
</CardGroup>

## Errors

Errors share a single envelope across every modality:

```json theme={"theme":{"light":"github-light","dark":"github-dark"}}
{
  "error": {
    "code": "invalid_input",
    "message": "Human-readable explanation.",
    "param": "model"
  }
}
```

`code` is the contract — branch on it, never on `message`. See the full table in [Errors](/gateway/errors).

## Rate limits

Every response carries `X-RateLimit-*` headers so you can back off without
guessing. When you exceed your quota, the gateway returns HTTP `429` with the
standard error envelope and a `Retry-After` header.