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

# Gemini Native generateContent API — API Reference

> POST /gw_llm_power/v1/models/{model}:generateContent — Use Gemini models natively with Google's protocol. Supports image and video analysis.

The Gemini Native API lets you interact with Gemini models using Google's native request format, including the `contents` and `generationConfig` structure. All Gemini models accessible through this endpoint support both image and video analysis, making it the right choice whenever you need multimodal capabilities with the full flexibility of the native Gemini protocol.

**Base URL:** `http://apillm.globalaiopc.com/gw_llm_power`

**Endpoints:**

* `POST /v1/models/{model}:generateContent` — Standard (non-streaming) response
* `POST /v1/models/{model}:streamGenerateContent` — Streaming response

## Authentication

Authenticate every request using the `Authorization` header with your API key:

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

## Supported Models

All Gemini models available through this endpoint support image and video analysis.

| Model                           | Variants                      |
| ------------------------------- | ----------------------------- |
| `gemini-2.5-flash-lite`         | Standard only                 |
| `gemini-2.5-pro`                | Standard, `-official`, `-low` |
| `gemini-3-flash-preview`        | Standard, `-official`, `-low` |
| `gemini-3.1-flash-lite-preview` | Standard only                 |
| `gemini-3.1-pro-preview`        | Standard, `-official`, `-low` |

### Model Suffix Reference

| Suffix      | Description               |
| ----------- | ------------------------- |
| *(none)*    | Standard / stable version |
| `-official` | Official version          |
| `-low`      | Budget version            |

## Request Parameters

<ParamField body="contents" type="array" required>
  The array of message content objects that make up the conversation. Each object typically contains a `role` and a `parts` array.
</ParamField>

<ParamField body="contents[].role" type="string">
  The role of the message author. Use `user` for human turns and `model` for prior model turns in multi-turn conversations.
</ParamField>

<ParamField body="contents[].parts" type="array" required>
  An array of content parts for the message. Typically contains `text` objects for plain text input. The native Gemini multimodal structure (inline images, video, etc.) is also supported.
</ParamField>

<ParamField body="contents[].parts[].text" type="string">
  The text content of the part.
</ParamField>

<ParamField body="generationConfig.temperature" type="number">
  Controls the randomness of the model's output. Lower values produce more focused, deterministic responses; higher values produce more creative output.
</ParamField>

<ParamField body="generationConfig.topP" type="number">
  Nucleus sampling parameter. The model considers only the tokens comprising the top `topP` probability mass.
</ParamField>

<ParamField body="generationConfig.maxOutputTokens" type="integer">
  The maximum number of tokens the model may generate in its response.
</ParamField>

<ParamField body="systemInstruction.parts[].text" type="string">
  An optional system prompt that sets the context and behavior for the model. Provide this as a text part within the `systemInstruction` object.
</ParamField>

## Response Fields

<ResponseField name="candidates[].content.parts[].text" type="string">
  The text content generated by the model.
</ResponseField>

<ResponseField name="candidates[].finishReason" type="string">
  The reason the model stopped generating. Common values include `STOP` (natural end) and `MAX_TOKENS` (token limit reached).
</ResponseField>

<ResponseField name="usageMetadata.promptTokenCount" type="integer">
  The number of tokens in the input contents and system instruction.
</ResponseField>

<ResponseField name="usageMetadata.candidatesTokenCount" type="integer">
  The number of tokens in the generated response candidates.
</ResponseField>

<ResponseField name="usageMetadata.totalTokenCount" type="integer">
  The total number of tokens used in the request (prompt + candidates).
</ResponseField>

## Code Examples

<CodeGroup>
  ```bash cURL theme={null}
  curl -X POST "http://apillm.globalaiopc.com/gw_llm_power/v1/models/gemini-3.1-pro-preview:generateContent" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "contents": [
        {
          "role": "user",
          "parts": [{"text": "Explain the history of artificial intelligence"}]
        }
      ],
      "generationConfig": {
        "temperature": 0.7,
        "maxOutputTokens": 1024
      }
    }'
  ```

  ```python Python theme={null}
  import requests

  response = requests.post(
      "http://apillm.globalaiopc.com/gw_llm_power/v1/models/gemini-3.1-pro-preview:generateContent",
      headers={
          "Authorization": "Bearer YOUR_API_KEY",
          "Content-Type": "application/json"
      },
      json={
          "contents": [
              {
                  "role": "user",
                  "parts": [{"text": "Explain the history of artificial intelligence"}]
              }
          ],
          "generationConfig": {
              "temperature": 0.7,
              "maxOutputTokens": 1024
          }
      }
  )
  print(response.json()["candidates"][0]["content"]["parts"][0]["text"])
  ```
</CodeGroup>

### Example Response

```json theme={null}
{
  "candidates": [
    {
      "content": {
        "parts": [
          {
            "text": "The history of artificial intelligence spans decades, progressing through three major phases: early rule-based systems in the 1950s–1980s, the rise of machine learning in the 1990s–2010s, and the current era of large language models and deep learning."
          }
        ],
        "role": "model"
      },
      "finishReason": "STOP",
      "index": 0
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 10,
    "candidatesTokenCount": 52,
    "totalTokenCount": 62
  }
}
```

<Note>
  To receive a streaming response, replace `:generateContent` with `:streamGenerateContent` in the request URL. The API will return a series of incremental response chunks in the native Gemini streaming format.
</Note>

<Tip>
  This endpoint follows the native Google Gemini protocol. For complete details on multimodal input, function calling, safety settings, and advanced generation configuration, refer to the [Google Gemini generateContent documentation](https://ai.google.dev/api/generate-content).
</Tip>