Gemini Native Format

Call Gemini models via the Google Gemini native API protocol. Fully compatible with Gemini's native request format, path structure, and response format.

Try it now

Test in Playground ↗ · View Model List ↗

Quick Start

Step 1: Get your API Key from the Console.

Step 2: Send a request:

cURLPythonNode.js

bash

curl -X POST "https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [{"text": "Hello, introduce yourself"}]
      }
    ]
  }'

python

import requests

url = "https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer sk-xxx"
}
payload = {
    "contents": [
        {
            "role": "user",
            "parts": [{"text": "Hello, introduce yourself"}]
        }
    ]
}

response = requests.post(url, json=payload, headers=headers)
result = response.json()
print(result["candidates"][0]["content"]["parts"][0]["text"])

javascript

const response = await fetch('https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    Authorization: 'Bearer sk-xxx',
  },
  body: JSON.stringify({
    contents: [
      {
        role: 'user',
        parts: [{ text: 'Hello, introduce yourself' }],
      },
    ],
  }),
})

const result = await response.json()
console.log(result.candidates[0].content.parts[0].text)

Step 3: Read the response from candidates[0].content.parts[0].text.

Endpoint

Item	Value
Method	POST
Path	`/v1beta/models/{model}:generateContent`
	`/v1beta/models/{model}:streamGenerateContent` (streaming)
Base URL	`https://open.dieyuyun.com`
Protocol	Google Gemini API

Path Parameters

Parameter	Type	Required	Description
model	string	Yes	Model identifier, e.g. `gemini-2.5-pro`
action	string	Yes	Operation type: `generateContent` or `streamGenerateContent`

Authentication

Uses Bearer Token authentication (shares the same API Key as the OpenAI-compatible endpoint):

http

Authorization: Bearer sk-xxx

TIP

The platform uses a unified API Key for all endpoints. No separate Google API Key is needed. Your API Key is securely forwarded to the upstream provider on the server side.

Supported Models

Model	Context Length	Description
gemini-2.5-pro	1M	Gemini flagship reasoning model
gemini-2.5-flash	1M	Gemini fast model
gemini-2.0-flash	1M	Gemini 2.0 fast model

TIP

See the full model list in the Console.

Request Parameters

Field	Type	Required	Default	Description
contents	array	Yes	—	Conversation content list, each with `role` and `parts`
systemInstruction	object	No	—	System instruction to define model behavior
generationConfig	object	No	—	Generation parameter configuration
safetySettings	array	No	—	Safety settings
tools	array	No	—	Tool definitions
toolConfig	object	No	—	Tool calling configuration

generationConfig Parameters

Parameter	Type	Default	Description
temperature	number	1.0	Sampling randomness, range 0~2
topP	number	0.95	Nucleus sampling parameter
topK	integer	40	Top-K sampling
maxOutputTokens	integer	—	Maximum output tokens
stopSequences	array	—	Strings that stop generation
responseMimeType	string	—	Response MIME type, e.g. `application/json`

contents Structure

json

{
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "Describe this image" },
        {
          "inline_data": {
            "mime_type": "image/jpeg",
            "data": "/9j/4AAQ..."
          }
        }
      ]
    }
  ]
}

Request Examples

With System Instruction and Generation Config

cURLPythonNode.js

bash

curl -X POST "https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "systemInstruction": {
      "parts": [{"text": "You are a professional translator. Translate to English."}]
    },
    "contents": [
      {
        "role": "user",
        "parts": [{"text": "Artificial intelligence is changing our way of life."}]
      }
    ],
    "generationConfig": {
      "temperature": 0.7,
      "maxOutputTokens": 1024
    }
  }'

python

import requests

url = "https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer sk-xxx"
}
payload = {
    "systemInstruction": {
        "parts": [{"text": "You are a professional translator. Translate to English."}]
    },
    "contents": [
        {
            "role": "user",
            "parts": [{"text": "Artificial intelligence is changing our way of life."}]
        }
    ],
    "generationConfig": {
        "temperature": 0.7,
        "maxOutputTokens": 1024
    }
}

response = requests.post(url, json=payload, headers=headers)
result = response.json()
print(result["candidates"][0]["content"]["parts"][0]["text"])

javascript

const response = await fetch('https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:generateContent', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    Authorization: 'Bearer sk-xxx',
  },
  body: JSON.stringify({
    systemInstruction: {
      parts: [{ text: 'You are a professional translator. Translate to English.' }],
    },
    contents: [
      {
        role: 'user',
        parts: [{ text: 'Artificial intelligence is changing our way of life.' }],
      },
    ],
    generationConfig: {
      temperature: 0.7,
      maxOutputTokens: 1024,
    },
  }),
})

const result = await response.json()
console.log(result.candidates[0].content.parts[0].text)

Streaming Output

Use the streamGenerateContent endpoint for streaming responses:

bash

curl -X POST "https://open.dieyuyun.com/v1beta/models/gemini-2.5-pro:streamGenerateContent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [{"text": "Write a poem about the ocean"}]
      }
    ]
  }'

Multimodal Input (Images)

json

{
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "What's in this image?" },
        {
          "inline_data": {
            "mime_type": "image/png",
            "data": "iVBORw0KGgo..."
          }
        }
      ]
    }
  ]
}

Function Calling

json

{
  "contents": [
    {
      "role": "user",
      "parts": [{ "text": "What's the weather like in Beijing today?" }]
    }
  ],
  "tools": [
    {
      "functionDeclarations": [
        {
          "name": "get_weather",
          "description": "Get weather information for a city",
          "parameters": {
            "type": "object",
            "properties": {
              "city": {
                "type": "string",
                "description": "City name"
              }
            },
            "required": ["city"]
          }
        }
      ]
    }
  ]
}

Response Format

Successful Response

json

{
  "candidates": [
    {
      "content": {
        "parts": [
          {
            "text": "Hello! I'm Gemini, a multimodal AI model developed by Google DeepMind. I can process text, images, code, and more."
          }
        ],
        "role": "model"
      },
      "finishReason": "STOP",
      "index": 0
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 6,
    "candidatesTokenCount": 42,
    "totalTokenCount": 48
  },
  "modelVersion": "gemini-2.5-pro"
}

Field Reference

Field	Description
candidates[].content.parts[].text	Response text content
candidates[].content.role	Role, `model` for model responses
candidates[].finishReason	Stop reason: `STOP`, `MAX_TOKENS`, `SAFETY`
candidates[].safetyRatings	Safety ratings
usageMetadata.promptTokenCount	Number of input tokens
usageMetadata.candidatesTokenCount	Number of output tokens
usageMetadata.totalTokenCount	Total token count
modelVersion	Actual model version used

Error Response

json

{
  "error": {
    "code": 400,
    "message": "Request contains an invalid argument.",
    "status": "INVALID_ARGUMENT"
  }
}

See Error Codes for details.

Compatibility

Feature	Wuliang AI	Google Gemini Native API
Path format	`/v1beta/models/{model}:action`	Same
Request body format	Gemini native JSON	Native protocol
generateContent	Supported	Native support
streamGenerateContent	Supported	Native support
Multimodal input	Supports base64	Supports base64 / file_data
Function calling	Supported	Native support
System instruction	Supports systemInstruction	Native support
Authentication	Bearer Token (platform API Key)	Google API Key
safetySettings	Pass-through support	Native support

TIP

The platform uses Gemini's native paths and request body format. Successful responses return the Gemini native format directly. The only difference is authentication -- the platform's unified API Key replaces the Google API Key.

Best Practices

Choose the right endpoint: Use generateContent for non-streaming and streamGenerateContent for streaming. The path structure is identical.
Leverage long context: Gemini supports up to 1M tokens of context, ideal for processing long documents, large codebases, and more.
Use systemInstruction: Define model behavior via system instructions. Note the Gemini format: {"parts": [{"text": "..."}]}.
Multimodal input: Supports base64-encoded images for image analysis, OCR, and other visual tasks.
JSON mode output: Force JSON output via generationConfig.responseMimeType = "application/json".

Rate Limits

See Rate Limits for details.

Chat Completions (Streaming) - Use the OpenAI-compatible protocol
Claude Messages - Use the Anthropic native protocol
OpenAI Responses - Use the OpenAI Responses API
Manage API Keys - Create and configure API keys

Gemini Native Format ​

Quick Start ​

Endpoint ​

Path Parameters ​

Authentication ​

Supported Models ​

Request Parameters ​

generationConfig Parameters ​

contents Structure ​

Request Examples ​

With System Instruction and Generation Config ​

Streaming Output ​

Multimodal Input (Images) ​

Function Calling ​

Response Format ​

Successful Response ​

Field Reference ​

Error Response ​

Compatibility ​

Best Practices ​

Rate Limits ​

Related Docs ​

Gemini Native Format

Quick Start

Endpoint

Path Parameters

Authentication

Supported Models

Request Parameters

generationConfig Parameters

contents Structure

Request Examples

With System Instruction and Generation Config

Streaming Output

Multimodal Input (Images)

Function Calling

Response Format

Successful Response

Field Reference

Error Response

Compatibility

Best Practices

Rate Limits

Related Docs