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

# Endpoint overview

> Reference APIONE protocol, model, image, and video endpoints.

All endpoints use HTTPS and authenticate with the same APIONE API Key.

| Method | Complete endpoint | Purpose |
| - | - | - |
| POST | `https://api.apione.ai/v1/chat/completions` | OpenAI Chat Completions |
| POST | `https://api.apione.ai/v1/responses` | OpenAI Responses |
| POST | `https://api.apione.ai/v1/messages` | Anthropic Messages |
| POST | `https://api.apione.ai/v1beta/models/{model}:generateContent` | Gemini non-streaming generation |
| POST | `https://api.apione.ai/v1beta/models/{model}:streamGenerateContent` | Gemini streaming generation |
| GET | `https://api.apione.ai/v1/models` | Query models available to the current key |
| GET | `https://api.apione.ai/v1/models/{model}/params` | Query model parameters and billing modes |
| POST | `https://api.apione.ai/v1/images/generations` | Image generation |
| POST | `https://api.apione.ai/v1/images/edits` | Image editing |
| POST | `https://api.apione.ai/v1/videos` | Create a video task |
| GET | `https://api.apione.ai/v1/videos/{id}` | Query video task status |

## Common request headers

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

Anthropic and Gemini native protocols also support their respective key headers. Send only one credential per request.

Save the actual request identifiers and field names returned in the response: `X-Request-ID`, `X-Maxx-Trace-Id`, or `X-Maxx-Request-ID`. Different interfaces do not guarantee returning all fields; record local business IDs separately and do not assume custom headers will be echoed back unchanged.

## Version paths

`/v1` and `/v1beta` are part of the interface version. When configuring an SDK, do not mistakenly enter the complete endpoint as the Base URL, and do not form `/v1/v1/...`.

## Find endpoints by task

### Text chat

Use Chat Completions for ordinary multi-turn conversations; use Responses for existing Responses workflows or Codex-like tools. The two cannot simply swap paths while reusing the exact same request body.

### Native protocols

Anthropic Messages and Gemini generateContent use their own message structures and authentication conventions. When choosing a native protocol, construct requests according to the corresponding page; do not send OpenAI's `messages` structure directly to the Gemini endpoint.

### Images and video

Image generation, image editing, and video generation are media interfaces. Query model parameters before submitting. Video is an asynchronous task: `POST /v1/videos` creates it, `GET /v1/videos/{id}` queries it; do not let a single creation request wait until generation finishes.

## General response handling

* Success responses: read the body, result object, or task ID according to the specific interface.
* Error responses: record both HTTP status and sanitized response body.
* Request tracking: save the actual returned request identifiers and field names to locate them in the console or support tickets.
* Dynamic capabilities: models, parameters, prices, and output formats are determined by the current key's query results.

## Pre-production checklist

```text theme={null}
[ ] Domain is api.apione.ai, not the console address
[ ] SDK Base URL and handwritten HTTP complete endpoints are not mixed up
[ ] /v1 or /v1beta appears only once
[ ] Key and model belong to the same APIONE account and available group
[ ] Timeout, retry, and logging strategies are set
[ ] Logs do not output the full API Key or sensitive input
```

Next: [Choose a protocol](https://apione.ai/docs/en/getting-started/protocol-selection.md), [Request tracking and retries](https://apione.ai/docs/en/api/request-reliability.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.