来源https://www.laoyingai.com/使用Python调用ChatGPT API主要依赖OpenAI官方库以下是完整的调用方法和代码示例。1. 安装与配置首先安装官方Python库bashpip install openai获取API密钥后推荐使用环境变量管理创建.env文件或直接设置环境变量pythonimport os from openai import OpenAI # 从环境变量读取密钥推荐 client OpenAI( api_keyos.environ.get(OPENAI_API_KEY) )安全提醒切勿将API密钥硬编码在代码中。2. 基础调用方法方法一Chat Completions API当前标准pythonfrom openai import OpenAI client OpenAI() completion client.chat.completions.create( modelgpt-4o, # 或 gpt-4-turbo, gpt-3.5-turbo messages[ {role: system, content: 你是一个专业的编程助手。}, {role: user, content: 如何在Python中实现快速排序} ], temperature0.7 # 控制随机性0-1之间 ) print(completion.choices[0].message.content)方法二Responses API较新的主要APIpythonresponse client.responses.create( modelgpt-4o, instructions你是一个友好的客服助手, input请问你们支持7天无理由退货吗 ) print(response.output_text)参数说明model模型名称如gpt-4o、gpt-4-turbo、gpt-3.5-turbomessages对话历史每个消息包含rolesystem/user/assistant和contenttemperature0-1之间值越低输出越确定越高越有创造性max_tokens限制响应最大长度3. 多轮对话管理多轮对话需要维护完整的消息历史pythondef run_conversation(): messages [ {role: system, content: 你是一个善于倾听的助手。} ] while True: user_input input(用户: ) if user_input.lower() quit: break # 添加用户消息 messages.append({role: user, content: user_input}) # 获取响应 completion client.chat.completions.create( modelgpt-4o, messagesmessages ) assistant_reply completion.choices[0].message.content print(f助手: {assistant_reply}) # 添加助手回复到历史 messages.append({role: assistant, content: assistant_reply})关键点每次调用都需要发送完整的历史记录否则模型会丢失上下文。4. 流式响应Streaming适用于聊天界面等需要实时输出的场景python# 同步流式 stream client.chat.completions.create( modelgpt-4o, messages[{role: user, content: 讲一个关于独角兽的短故事}], streamTrue ) for chunk in stream: if chunk.choices[0].delta.content: print(chunk.choices[0].delta.content, end, flushTrue)python# 异步流式 from openai import AsyncOpenAI import asyncio async def stream_response(): client AsyncOpenAI() stream await client.chat.completions.create( modelgpt-4o, messages[{role: user, content: 你好}], streamTrue ) async for chunk in stream: if chunk.choices[0].delta.content: print(chunk.choices[0].delta.content, end) asyncio.run(stream_response())5. 异步调用提升并发性能适用于高吞吐场景pythonimport asyncio from openai import AsyncOpenAI async def main(): client AsyncOpenAI() response await client.chat.completions.create( modelgpt-4o, messages[{role: user, content: 解释什么是异步编程}] ) print(response.choices[0].message.content) asyncio.run(main())6. 函数调用Function Calling让模型调用自定义函数获取实时数据pythonimport json from openai import OpenAI # 定义工具函数 def get_weather(latitude: float, longitude: float) - str: 获取指定坐标的天气 # 实际应调用真实API return f{latitude}, {longitude} 位置的温度为15°C # 定义工具描述 tools [{ type: function, function: { name: get_weather, description: 获取指定经纬度的当前温度, parameters: { type: object, properties: { latitude: {type: number}, longitude: {type: number} }, required: [latitude, longitude] } } }] client OpenAI() messages [{role: user, content: 纽约现在的天气怎么样}] completion client.chat.completions.create( modelgpt-4o, messagesmessages, toolstools ) # 处理工具调用 if completion.choices[0].finish_reason tool_calls: tool_call completion.choices[0].message.tool_calls[0] arguments json.loads(tool_call.function.arguments) # 执行函数 result get_weather(**arguments) # 将结果返回给模型 messages.append(completion.choices[0].message) messages.append({ role: tool, tool_call_id: tool_call.id, content: result }) # 获取最终回复 final_response client.chat.completions.create( modelgpt-4o, messagesmessages ) print(final_response.choices[0].message.content)7. 错误处理与重试pythonfrom openai import OpenAI, APIError, APIConnectionError, RateLimitError import time def call_with_retry(max_retries3): for attempt in range(max_retries): try: completion client.chat.completions.create( modelgpt-4o, messages[{role: user, content: Hello}] ) return completion.choices[0].message.content except RateLimitError: wait_time min(2 ** attempt 0.5, 10) print(f速率限制等待 {wait_time} 秒...) time.sleep(wait_time) except APIConnectionError: print(连接错误重试中...) time.sleep(1) except APIError as e: print(fAPI错误: {e}) break raise Exception(所有重试尝试失败)8. 结构化输出JSON模式pythoncompletion client.chat.completions.create( modelgpt-4o, messages[ {role: system, content: 从文本中提取姓名和城市以JSON格式返回。}, {role: user, content: 张三住在北京是一名工程师。} ], response_format{type: json_object} ) print(completion.choices[0].message.content) # 输出{name: 张三, city: 北京}注意事项成本控制注意max_tokens限制响应token比输入token成本更高速率限制免费层每分钟3次调用生产环境需升级计划API版本确保使用最新SDKv1.0旧版openai.Completion已废弃来源 https://www.laoyingai.com/