Blame for API.md

# Gidora API Documentation

**Model:** Gidora  
**Model ID:** `Gidora-v03.01`  
**Version:** `v03.01`  
**Developer:** GidoraCo

## 1. Overview

Gidora is an AI language model developed by GidoraCo.

The model provides text-based capabilities, including question answering, text generation, text analysis, summarization, translation, and information extraction.

Gidora supports Persian and other major languages and is optimized for Persian-language tasks.

The API follows an OpenAI-compatible chat completion format.

## 2. API Access and Security

Gidora is hosted on dedicated infrastructure and is accessible through secure, dedicated API gateways.

The API uses Bearer Token authentication.

The Base URL and API endpoints are not publicly disclosed. Authorized users and evaluation teams can obtain connection details through secure communication channels.

Example authentication header:

```http
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json
```

**Security notice:** Never publish actual API keys, private URLs, endpoints, or infrastructure credentials in this repository.

## 3. Model Identifier

Use the following model identifier in API request bodies:

```text
Gidora-v03.01
```

## 4. Chat Completion Request

The following example illustrates the request body format provided for Gidora.

```json
{
  "model": "Gidora-v03.01",
  "messages": [
    {
      "role": "system",
      "content": "You are a helpful assistant."
    },
    {
      "role": "user",
      "content": "سلام، چطور می‌توانم در مورد یادگیری زبان‌های برنامه‌نویسی به شما سوال بپرسم؟"
    }
  ],
  "temperature": 0.7,
  "max_tokens": 512,
  "top_p": 0.9
}
```

### Request Parameters

| Parameter | Description |
|---|---|
| `model` | Model identifier |
| `messages` | Conversation messages |
| `role` | Message role, such as system, user, or assistant |
| `content` | Text content of a message |
| `temperature` | Controls response randomness |
| `max_tokens` | Maximum number of generated tokens |
| `top_p` | Nucleus sampling parameter |

The supported values and limits of optional parameters should be confirmed with the API provider.

## 5. Chat Completion Response

The following example illustrates the response body format provided for Gidora.

```json
{
  "id": "req_123456789",
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "سلام! من می‌توانم در آموزش زبان‌های مختلف برنامه‌نویسی، توضیح مفاهیم پایه، رفع خطاهای کدنویسی و ارائه نمونه‌های عملی به شما کمک کنم. چه موضوعی مد نظرتان است؟"
      },
      "finish_reason": "stop",
      "index": 0
    }
  ],
  "usage": {
    "prompt_tokens": 45,
    "completion_tokens": 80,
    "total_tokens": 125
  }
}
```

### Response Fields

| Field | Description |
|---|---|
| `id` | Request identifier |
| `choices` | Generated completion choices |
| `message` | Generated assistant message |
| `role` | Role of the message |
| `content` | Generated text |
| `finish_reason` | Reason generation finished |
| `index` | Choice index |
| `usage` | Token usage information |
| `prompt_tokens` | Number of input tokens |
| `completion_tokens` | Number of generated tokens |
| `total_tokens` | Total token usage |

## 6. Multi-turn Conversations

Gidora supports multi-turn conversations.

The `messages` array can contain multiple messages with different roles, including:

- `system`: Defines the assistant's behavior.
- `user`: Contains the user's message.
- `assistant`: Contains a previous assistant response.

Conversation history can be included in subsequent requests to maintain context.

## 7. Streaming

Streaming responses are supported by the Gidora API.

Detailed streaming request parameters and response event formats are not included in this public document.

Authorized users can obtain additional integration details through the appropriate secure channel.

## 8. Rate Limits

**Reported rate limit:** 300 requests per minute (RPM).

The applicable rate-limit scope and allocation policy should be confirmed with GidoraCo.

## 9. API Compatibility

Gidora provides an OpenAI-compatible API interface.

The examples in this document describe the chat request and response structure. Connection details are provided separately to authorized users.

## 10. Model Availability

Gidora is hosted on GidoraCo's dedicated infrastructure.

Model weights are proprietary and are not publicly distributed through this repository.

Access is provided through the dedicated API service.

## 11. Evaluation Access

Restricted API access for independent evaluation can be arranged through GidoraCo.

Evaluation teams should request credentials and connection details through a secure communication channel.

Actual API credentials and private infrastructure information must never be included in public documentation.

## 12. Contact

**Developer:** GidoraCo  
**Website:** https://ai.gidora.com