Gemini Native Format

Gemini native interface supports both text and image generation. This interface allows you to directly call Gemini's native capabilities, including the latest image generation features.

Interface Details

Endpoint: POST /v1beta/models/{model}:generateContent

Description: Generates content based on prompts and configurations. Supports multi-modal input (text, images) and multi-modal output (text, images).

Authentication: Bearer Token

Authorization: Bearer YOUR_API_TOKEN

---

Request Parameters

Path Parameters

Parameter	Type	Required	Description	Example
model	string	Yes	Model Name	gemini-2.0-flash-exp

Header Parameters

Parameter	Type	Required	Description	Example
Authorization	string	Yes	Bearer Token Auth	Bearer sk-xxx...
Content-Type	string	Yes	Content Type	application/json

Body Parameters

Parameter	Type	Required	Description
contents	array[object]	Yes	List of dialogue contents
contents[].role	string	No	Role (user, model)
contents[].parts	array[object]	Yes	Content parts
contents[].parts[].text	string	No	Text prompt
generationConfig	object	No	Generation config options
generationConfig.responseModalities	array[string]	No	Response modalities (TEXT, IMAGE)
generationConfig.imageConfig	object	No	Image generation configuration
generationConfig.imageConfig.aspectRatio	string	No	Aspect ratio (1:1, 16:9, 9:16, etc.)
generationConfig.imageConfig.imageSize	string	No	Output resolution tier, see "Image Resolution" below

Image Resolution

Which imageSize tiers are available depends on the model. Omit the field to let the model apply its 1K default.

Tier	Value	Supported models
0.5K	512	gemini-3.1-flash-image family only
1K	1K	All image models (default)
2K	2K	gemini-3-pro-image family
4K	4K	gemini-3-pro-image family

The wire value for the 0.5K tier is 512, not 0.5K. So that callers who write the tier by its documented name still work, the gateway rewrites 0.5K (case-insensitive), 512P and 512PX to 512, and all of these spellings are accepted.

Requesting the 0.5K tier on a model that does not support it makes the gateway drop the field and fall back to the model's 1K default rather than fail the request. A tier the gateway does not recognise is forwarded untouched for the model to validate.

---

Response Parameters

Parameter	Type	Description
candidates	array[object]	List of generation candidates
candidates[].content	object	Content object
candidates[].content.parts	array[object]	Content parts, containing text or generated image data
candidates[].finishReason	string	Reason for completion
usageMetadata	object	Usage statistics
usageMetadata.promptTokenCount	integer	Prompt token count
usageMetadata.candidatesTokenCount	integer	Generated token count
usageMetadata.totalTokenCount	integer	Total token count

---

Code Examples

Curl Example (Image Generation)

curl -X POST "https://api.ezmodel.cloud/v1beta/models/gemini-2.0-flash-exp:generateContent" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "parts": [
          {
            "text": "A cyberpunk style fox running in a forest"
          }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {
        "aspectRatio": "16:9",
        "imageSize": "1K"
      }
    }
  }'

Response Example

{
  "candidates": [
    {
      "content": {
        "role": "model",
        "parts": [
          {
            "inlineData": {
              "mimeType": "image/png",
              "data": "iVBORw0KGgoAAA..."
            }
          }
        ]
      },
      "finishReason": "STOP"
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 15,
    "candidatesTokenCount": 0,
    "totalTokenCount": 15
  }
}