前缀续写 - AI 智算产品文档

对话前缀续写是指在给定一段对话的开头部分（即对话前缀）的基础上，让大语言模型继续生成后续的对话内容。

支持的模型

注意事项

用户需提前在 API 密钥管理中创建和获取 API KEY。
API 调用时，用户需将 messages 数组列表里最后一条消息中的 role 设置为 assistant。

代码示例

Curl

curl -X POST https://openapi.coreshub.cn/v1/chat/completions \
-H "Authorization: Bearer sk-xxxxxxxxxx" \
-H "Content-Type: application/json" \
-d '{
    "model": "DeepSeek-V3",
    "messages": [{
        "role": "user",
        "content": "请对“叮铃铃，下课铃响了”这句话进行续写，用一句话来描述校园里突然生机和活力的场景"
    },
    {
        "role": "assistant",
        "content": "叮铃铃，下课铃响了"
    }]
}'

Python

示例一：

示例需求如下：

设置 assistant 开头消息为 "```python\n"，用于强制模型输出 Python 代码。
设置 stop 参数为 ['```'] 用于避免模型的额外解释。

from openai import OpenAI

client = OpenAI(api_key='sk-xxxxxxxxxx', base_url='https://openapi.coreshub.cn/v1')

messages = [
    {"role": "user", "content": "Please write quick sort code"},
    {"role": "assistant", "content": "```python\n"}
]
response = client.chat.completions.create(
    model="DeepSeek-V3",
    messages=messages,
    stop=["```"],
)
print(response.choices[0].message.content)

示例二：

import os
from openai import OpenAI

client = OpenAI(api_key='sk-xxxxxxxxxx', base_url='https://openapi.coreshub.cn/v1')

completion = client.chat.completions.create(
    model="DeepSeek-V3",
    messages=[{
        "role": "user",
        "content": "请对“叮铃铃，下课铃响了”这句话进行续写，用一句话来描述校园里突然生机和活力的场景"
    },
    {
        "role": "assistant",
        "content": "叮铃铃，下课铃响了"
    }]
    )
print(completion.choices[0].message.content)

返回结果

Curl

{
    "id": "902d63b5f10940e08f50dde32224b15d",
    "object": "chat.completion",
    "created": 1740987764,
    "model": "DeepSeek-V3",
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "，原本安静的校园瞬间沸腾起来，学生们如欢快的小鸟般涌出教室，操场上顿时充满了欢声笑语和青春的气息。",
                "tool_calls": null
            },
            "logprobs": null,
            "finish_reason": "stop",
            "matched_stop": 1
        }
    ],
    "usage": {
        "prompt_tokens": 37,
        "total_tokens": 67,
        "completion_tokens": 30,
        "prompt_tokens_details": null
    }
}

Python

示例一：
示例二：