Gemini聊天对话

POST

/models/gemini-3-pro-preview:streamGenerateContent

使用方式与 Gemini 官方保持一致，该接口只支持请求 Gemini 系列模型，更多配置请看模型官方文档 https://ai.google.dev/gemini-api/docs?hl=zh-cn。
非流请求是：/v1beta/models/gemini-3-pro-preview:generateContent

请求参数

Query 参数

Header 参数

Body 参数application/json必填

示例

{
    "contents": [
        {
            "parts": [
                {
                    "text": "你是谁"
                }
            ]
        }
    ]
}

请求示例代码

Shell

JavaScript

Java

Swift

PHP

Python

HTTP

Objective-C

Ruby

OCaml

Dart

curl --location '/models/gemini-3-pro-preview:streamGenerateContent?key=sk-xxx&alt=sse' \
--header 'Authorization;' \
--header 'Content-Type: application/json' \
--data '{
    "contents": [
        {
            "parts": [
                {
                    "text": "你是谁"
                }
            ]
        }
    ]
}'

返回响应

🟢200成功

application/json

Bodyapplication/json

示例

{
    "candidates": [
        {
            "content": {
                "role": "model",
                "parts": [
                    {
                        "text": "**Defining My Identity**\n\nI've just finished breaking down the user's Chinese prompt, \"你是谁?\" It seems straightforward: I need to introduce myself. Now I'm considering different ways to answer; a concise response, versus a more detailed description of my nature. Currently, I'm leaning toward the detailed approach. The system prompt is influencing me, too, and the key is a smooth, accurate translation into Chinese.\n\n\n",
                        "thought": true
                    },
                    {
                        "text": "我是 Gemini，由 Google 开发的大型语言模型。",
                        "thoughtSignature": "CvQHAePx/16Kxxxxxxxxxxxxxxxxxxxxx"
                    }
                ]
            },
            "finishReason": "STOP"
        }
    ],
    "usageMetadata": {
        "promptTokenCount": 2,
        "candidatesTokenCount": 11,
        "totalTokenCount": 270,
        "trafficType": "ON_DEMAND",
        "promptTokensDetails": [
            {
                "modality": "TEXT",
                "tokenCount": 2
            }
        ],
        "candidatesTokensDetails": [
            {
                "modality": "TEXT",
                "tokenCount": 11
            }
        ],
        "thoughtsTokenCount": 257
    },
    "modelVersion": "gemini-3-pro-preview",
    "createTime": "2025-11-27T15:45:15.490458Z",
    "responseId": "C3Ioadr3HcqE998P26jViQY"
}

修改于 2026-05-07 12:34:41

Gemini媒体识别

实时语音WebSocket