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

# API manual overview

> OmniMux gateway protocols, auth, and async task behavior.

Call the gateway with a **Bearer token**. Production Base URL:

```text theme={null}
https://api.omnimux.ai
```

Typical OpenAI-compatible SDK value:

```text theme={null}
https://api.omnimux.ai/v1
```

Console: [omnimux.ai/dashboard](https://omnimux.ai/dashboard) · Docs: [docs.omnimux.ai](https://docs.omnimux.ai)

## Authentication

```http theme={null}
Authorization: Bearer sk-xxxxxxxx
Content-Type: application/json
```

Create tokens under **Tokens / API Keys** in the console. See [Authentication](/en/guides/authentication).

## Protocols and capabilities

| Capability        | Method         | Path                                     | Notes                         |
| ----------------- | -------------- | ---------------------------------------- | ----------------------------- |
| Chat Completions  | `POST`         | `/v1/chat/completions`                   | OpenAI Chat-compatible        |
| Responses         | `POST`         | `/v1/responses`                          | OpenAI Responses              |
| Claude Messages   | `POST`         | `/v1/messages`                           | Anthropic Messages-compatible |
| Gemini            | `POST`         | `/v1beta/models/{model}:generateContent` | Gemini-native style           |
| List models       | `GET`          | `/v1/models`                             | OpenAI-style list             |
| Gemini model list | `GET`          | `/v1beta/models`                         | Gemini-style list             |
| Images            | `POST`         | `/v1/images/generations`                 | Often async                   |
| Video             | `POST` / `GET` | `/v1/video/generations`, `/v1/videos`, … | Async create + query          |

Sidebar endpoint pages are generated from `openapi/relay.json` and support try-it when playground is enabled.

## Sync vs async

| Mode  | Typical APIs                         | Behavior                                             |
| ----- | ------------------------------------ | ---------------------------------------------------- |
| Sync  | Chat, Messages, Responses, most text | Response body is the model result                    |
| Async | Images, video, …                     | Returns `task_id` first; poll for status and results |

Field names follow each endpoint schema. Image/video URLs may expire — download and store them promptly in production.

## Model IDs

Every request must use a `model` (or Gemini path `{model}`) enabled for your account:

```bash theme={null}
curl https://api.omnimux.ai/v1/models \
  -H "Authorization: Bearer $OMNIMUX_API_KEY"
```

Example names in docs are not proof of availability. See [Models](/en/guides/models).

## Errors

| HTTP         | Likely cause                                              |
| ------------ | --------------------------------------------------------- |
| `401`        | Missing or invalid Bearer token                           |
| `403`        | Model not allowed on token, or IP restricted              |
| `404`        | Wrong path, or extra/missing `/v1` relative to the client |
| `429`        | Rate limited                                              |
| Quota errors | Account or token quota exhausted                          |

## Related

* [Quickstart](/en/quickstart)
* [Configure Base URL](/en/guides/base-url)
* [Integration guide](/en/integration-guide/overview)

<Tip>
  Unimplemented placeholder ops are omitted from the public OpenAPI. If a model appears in the console but not here, trust live `GET /v1/models` and channel config.
</Tip>