大模型API接入指南

1

pip install openai requests dashscope volcengine-python-sdk

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

import os
from openai import OpenAI

# 配置（建议将密钥存入环境变量，避免硬编码）
# 方式1：直接设置环境变量（Windows：set OPENAI_API_KEY=你的密钥；Mac/Linux：export OPENAI_API_KEY=你的密钥）
client = OpenAI(
    api_key=os.getenv("OPENAI_API_KEY") or "你的OpenAI密钥",
    base_url="https://api.openai.com/v1"  # 替换为Groq地址：https://api.groq.com/openai/v1
)

def openai_chat_demo():
    """OpenAI 聊天接口示例"""
    try:
        response = client.chat.completions.create(
            model="gpt-4o-mini",  # 可选模型：gpt-5.2、llama-3-70b（Groq）
            messages=[
                {"role": "system", "content": "你是一个简洁的助手"},
                {"role": "user", "content": "请简要介绍Python的优势"}
            ],
            temperature=0.7,
            max_tokens=500
        )
        # 提取回复内容
        print("OpenAI回复：\n", response.choices[0].message.content)
    except Exception as e:
        print("调用失败：", str(e))

if __name__ == "__main__":
    openai_chat_demo()

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

import os
from openai import OpenAI

# Kimi 兼容OpenAI格式，配置专属base_url
client = OpenAI(
    api_key=os.getenv("KIMI_API_KEY") or "你的Kimi密钥",
    base_url="https://api.moonshot.cn/v1"
)

def kimi_chat_demo():
    """Kimi 聊天接口示例"""
    try:
        response = client.chat.completions.create(
            model="moonshot-v4",  # 可选模型：moonshot-v4-32k、moonshot-v3
            messages=[
                {"role": "system", "content": "你是一个专业的技术助手，回答简洁易懂"},
                {"role": "user", "content": "请简要说明本地部署Ollama的步骤"}
            ],
            temperature=0.6,
            max_tokens=1000
        )
        print("Kimi回复：\n", response.choices[0].message.content)
    except Exception as e:
        print("调用失败：", str(e))

if __name__ == "__main__":
    kimi_chat_demo()

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

import os
import dashscope
from dashscope import Generation

def dashscope_chat_demo():
    """阿里云百炼 聊天接口示例"""
    # 配置密钥
    dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") or "你的阿里云百炼密钥"
    
    try:
        response = Generation.call(
            model="qwen-turbo",  # 可选模型：qwen-max、qwen-omni-realtime
            messages=[
                {"role": "system", "content": "你是一个简洁的助手"},
                {"role": "user", "content": "请简要介绍阿里云百炼的优势"}
            ],
            result_format="message",
            temperature=0.7,
            max_tokens=500
        )
        # 提取回复内容
        if response.status_code == 200:
            print("阿里云百炼回复：\n", response.output.choices[0].message.content)
        else:
            print("调用失败：", response.code, response.message)
    except Exception as e:
        print("调用失败：", str(e))

if __name__ == "__main__":
    dashscope_chat_demo()

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

from openai import OpenAI

# Ollama 本地接口，兼容OpenAI格式
client = OpenAI(
    api_key="ollama",  # 占位符，无需真实密钥
    base_url="http://localhost:11434/v1"
)

def ollama_chat_demo():
    """Ollama 本地聊天接口示例（需先运行：ollama run llama3）"""
    try:
        response = client.chat.completions.create(
            model="llama3",  # 本地拉取的模型名称（可替换为mistral、qwen等）
            messages=[
                {"role": "system", "content": "你是一个简洁的本地助手"},
                {"role": "user", "content": "请简要介绍本地部署大模型的优势"}
            ],
            temperature=0.7,
            max_tokens=500
        )
        print("Ollama本地回复：\n", response.choices[0].message.content)
    except Exception as e:
        print("调用失败：", str(e), "（请确认Ollama已启动并拉取了llama3模型）")

if __name__ == "__main__":
    ollama_chat_demo()

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.getenv("TUZI_API_KEY") or "你的兔子API密钥",
    base_url="https://api.tu-zi.com/v1"
)

def tuzi_chat_demo():
    """兔子API 聚合聊天接口示例"""
    try:
        response = client.chat.completions.create(
            model="gpt-4o",  # 可选国内模型：kimi-moonshot-v4、glm-4
            messages=[
                {"role": "user", "content": "请简要说明聚合API的优势"}
            ],
            temperature=0.7,
            max_tokens=500
        )
        print("兔子API回复：\n", response.choices[0].message.content)
    except Exception as e:
        print("调用失败：", str(e))

if __name__ == "__main__":
    tuzi_chat_demo()