> ## Documentation Index
> Fetch the complete documentation index at: https://docs-vip.apigo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI 多模态对话示例

> OpenAI 图文混合输入示例。

## 推荐 endpoint

* [OpenAI /v1/responses](/api-reference/endpoints/openai/responses)

## 最小请求

```json theme={null}
{
  "model": "gpt-4.1",
  "input": [
    {
      "role": "user",
      "content": [
        { "type": "input_text", "text": "描述这张图里的界面结构。" },
        { "type": "input_image", "image_url": "https://example.com/ui.png" }
      ]
    }
  ]
}
```

## cURL 示例

```bash theme={null}
curl https://api-vip.apigo.ai/v1/responses \
  -H "Authorization: Bearer $YOUR API KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4.1",
    "input": [
      {
        "role": "user",
        "content": [
          { "type": "input_text", "text": "描述这张图里的界面结构。" },
          { "type": "input_image", "image_url": "https://example.com/ui.png" }
        ]
      }
    ]
  }'
```

## Python 示例

```python theme={null}
from openai import OpenAI

client = OpenAI(
    base_url="https://api-vip.apigo.ai/v1",
    api_key="<YOUR API KEY>",
)

response = client.responses.create(
    model="gpt-4.1",
    input=[
        {
            "role": "user",
            "content": [
                {"type": "input_text", "text": "描述这张图里的界面结构。"},
                {"type": "input_image", "image_url": "https://example.com/ui.png"},
            ],
        }
    ],
)

print(response.output_text)
```

## Node.js 示例

```js theme={null}
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api-vip.apigo.ai/v1",
  apiKey: process.env.YOUR API KEY,
});

const response = await client.responses.create({
  model: "gpt-4.1",
  input: [
    {
      role: "user",
      content: [
        { type: "input_text", text: "描述这张图里的界面结构。" },
        { type: "input_image", image_url: "https://example.com/ui.png" }
      ]
    }
  ]
});

console.log(response.output_text);
```

## 最佳实践

* 图片可以走 URL、base64 或 file ID，复用素材优先 file ID
* 多张图片时在服务端统一排序，不要依赖前端隐式顺序
* 大图和 OCR 场景单独评估 token 成本