Gemini 聊天 (含多模態)

通過統一端點呼叫 Gemini 2.x / 3.x 全系, 支援文本 + 影像多模態輸入與 thinking

POST /v1/chat/completions

鉴权: {'type': 'bearer', 'prefix': 'sk-', 'description': 'API Key, 使用 `Authorization: Bearer sk-xxx` 鉴权'}

Gemini 模型完全相容 OpenAI Chat Completions 協議。本頁展示 Gemini 特有能力: - **多模態內容陣列**: `content` 可以是 `[{type:"text",text:...},{type:"image_url",image_url:{url:...}}]` - **Thinking / 推理**: 2.5 Pro / 3.1 Pro Preview 可通過 `reasoning_effort: "low"|"medium"|"high"` 開啟深度推理 - **長上下文**: 2.5 Pro 支援 200K tokens, Flash 128K+ - **結構化輸出**: `response_format: {type:"json_object"}` 或傳 `response_mime_type`

请求体

`model`	string	required	Gemini 模型 ID, 如 `gemini-2.5-pro` / `gemini-2.5-flash` / `gemini-2.5-flash-lite` / `gemini-3.1-pro-preview` / `gemini-3.1-flash-lite-preview`
`messages`	array	required	多模態訊息陣列。`content` 可為字串或混合陣列
`role`	string	required
`content`	string	required	文本字串, 或內容塊陣列 `[{"type":"text","text":"..."},{"type":"image_url","image_url":{"url":"https://..."}}]`
`temperature`	number
`top_p`	number
`top_k`	integer		(Gemini 特有) Top-K 取樣, 建議不與 top_p 同時使用
`max_tokens`	integer
`stream`	boolean
`reasoning_effort`	string		(僅 2.5 Pro / 3.1 Pro Preview) 思考深度, high 更準但更慢
`response_format`	object		如 `{"type":"json_object"}` 強制 JSON 輸出
`stop`	array

响应

200 — Gemini 生成結果 (OpenAI 格式)

调用示例

curl https://api.router.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-2.5-pro",
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "text", "text": "这张图里是什么?"},
          {"type": "image_url", "image_url": {"url": "https://example.com/cat.jpg"}}
        ]
      }
    ],
    "reasoning_effort": "medium"
  }'

API 文件