/v1/chat/completions

基础地址：https://www.right.codes/draw

POST /v1/chat/completions

这个接口兼容 OpenAI 的聊天格式。纯文本、带图提问都能直接用。

统一鉴权头：Authorization: Bearer sk-xxxxx

请求参数

字段	必填	说明
model	是	支持所有模型
stream	是	true / false
messages	是	对话数组

messages[].content 可以直接是字符串，也可以按下面这种方式把图片一起塞进去：

[
  {
    "role": "user",
    "content": [
      {
        "type": "text",
        "text": "这张图片内容是什么"
      },
      {
        "type": "image_url",
        "image_url": {
          "url": "https://xxxxxxx.png"
        }
      }
    ]
  }
]

文字提问

{
  "model": "gemini-3.1-pro",
  "stream": false,
  "messages": [
    {
      "role": "user",
      "content": "你好"
    }
  ]
}

传图提问

{
  "model": "gemini-3.1-pro",
  "stream": false,
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "这张图片内容是什么"
        },
        {
          "type": "image_url",
          "image_url": {
            "url": "https://xxxxxxx.png"
          }
        }
      ]
    }
  ]
}

返回

JSON

返回里常用的是这些字段：

• choices[0].message.content
• usage.prompt_tokens
• usage.completion_tokens
• usage.total_tokens

{
  "id": "1-2ede12b5-77cc-48f9-b1d0-7ae35ee8d444",
  "object": "",
  "created": 1777897048,
  "model": "gemini-3.1-pro",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "你好！请问有什么我可以帮您的吗？"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 2,
    "completion_tokens": 261,
    "total_tokens": 263
  },
  "system_fingerprint": ""
}

Stream

stream=true 时，服务端会按 SSE 分片返回，正文里看 choices[0].delta.content。最后一个 chunk 会带上 usage。

data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"你好！","role":"assistant"},"finish_reason":null}]}

data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"请问有什么我可以帮你的吗？","role":"assistant"},"finish_reason":"stop"}]}

content_filter_results 也会跟着返回，结构和 OpenAI 兼容。