Skip to main content
POST
/
v1
/
models
/
{model}
:generateContent
Generate content (Gemini)
curl --request POST \
  --url https://api.dgrid.ai/v1/models/{model}:generateContent \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "contents": [
    {
      "role": "user",
      "parts": [
        {
          "text": "Hello from DGrid."
        }
      ]
    }
  ]
}
'
{
  "candidates": [
    {
      "content": {
        "role": "<string>",
        "parts": [
          {}
        ]
      },
      "finishReason": "<string>",
      "safetyRatings": [
        {}
      ]
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 123,
    "candidatesTokenCount": 123,
    "totalTokenCount": 123
  }
}
This page uses the same generateContent operation as Generate content (Gemini), with the playground above pre-filled for plain text chat. The notes below describe the Gemini-native multimodal fields you can add to contents[].parts to analyze images, audio, video, or mixed media in a single request.
Each part can carry inline data (base64-encoded bytes plus a MIME type) alongside text instructions, letting the model reason across modalities in one call.

​
Gemini-native request fields

The generic contents and generationConfig fields shown in the playground accept the following nested shape for multimodal recognition:
FieldTypeRequiredDescription
contents[].rolestringNoRole of the turn, e.g. user.
contents[].partsarrayYesOrdered list of content parts (text and/or inline media).
contents[].parts[].textstringNoText instruction or question for the model.
contents[].parts[].inlineDataobjectNoInline media payload for image, audio, or video understanding.
contents[].parts[].inlineData.mimeTypestringNoMIME type of the inline data, e.g. image/jpeg, audio/mp3, video/mp4.
contents[].parts[].inlineData.datastringNoBase64-encoded media bytes.
You can mix multiple parts in a single turn — for example a text part with an instruction followed by one or more inlineData parts containing the media to analyze.

​
Example: analyzing an image

{
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "Describe what is happening in this image." },
        {
          "inlineData": {
            "mimeType": "image/jpeg",
            "data": "<base64-encoded-image-bytes>"
          }
        }
      ]
    }
  ]
}

​
Response fields

The response follows the standard generateContent shape. The fields most relevant to media recognition are:
​
candidates
array
Candidate responses returned by the model.
​
usageMetadata
object
Token accounting metadata, including promptTokenCount, candidatesTokenCount, and totalTokenCount. Inline media (images, audio, video) consumes prompt tokens in addition to any text parts.

​
Example response

200
{
  "candidates": [
    {
      "content": {
        "role": "model",
        "parts": [
          { "text": "The image shows a golden retriever sitting on a grassy lawn." }
        ]
      },
      "finishReason": "STOP",
      "safetyRatings": []
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 264,
    "candidatesTokenCount": 18,
    "totalTokenCount": 282
  }
}

Authorizations

​
Authorization
string
header
required

Your DGrid API key. All endpoints use Authorization: Bearer <DGRID_API_KEY>.

Path Parameters

​
model
string
required

Target model ID, such as gemini-1.5-pro.

Body

application/json
​
contents
object[]

Input content array with role and parts.

​
generationConfig
object

Generation configuration.

Response

Generated content candidates.

​
candidates
object[]

Candidate responses returned by the model.

​
usageMetadata
object

Token accounting metadata.