调用OpenAI gpt-3.5-turbo API遇InvalidRequestError求助

核心问题分析 #

• 用错了API接口：gpt-3.5-turbo是聊天模型，必须用openai.ChatCompletion.create()接口，你之前用的openai.Completion.create()是给文本补全模型（比如text-davinci-003）设计的，不支持messages参数。
• 代码变量名错误：循环遍历的是item，但构建请求时误用了未定义的prompt变量。

修正后的完整代码 #

import openai
import os

# 别硬编码API密钥，建议用环境变量或Colab Secrets存储
openai.api_key = os.getenv("OPENAI_API_KEY")  # 替换成你的密钥获取方式

def get_response(prompts: list, model="gpt-3.5-turbo"):
    responses = []
    for item in prompts:
        response = openai.ChatCompletion.create(
            model=model,
            messages=[{"role": "user", "content": item}],  # 这里用item而非prompt
            temperature=0,
            max_tokens=20,
            top_p=1,
            frequency_penalty=0,
            presence_penalty=0
        )
        responses.append(response['choices'][0]['message']['content'].strip())
    return responses

# 测试调用
prompts = ['What are your functionalities?', 'what is the best name for an ice-cream shop?', 'who won the premier league last year?']
responses = get_response(prompts)
for idx, res in enumerate(responses):
    print(f"Prompt {idx+1}: {prompts[idx]}")
    print(f"Response: {res}\n")

额外优化建议 #

• 密钥安全：绝对不要把API密钥直接写在代码里，Colab可以用左侧栏的「Secrets」功能存储，再通过os.getenv读取。
• 异常处理：加上try-except块应对网络错误、配额不足等问题，避免程序直接崩溃：

def get_response(prompts: list, model="gpt-3.5-turbo"):
    responses = []
    for item in prompts:
        try:
            response = openai.ChatCompletion.create(
                model=model,
                messages=[{"role": "user", "content": item}],
                temperature=0,
                max_tokens=20,
                top_p=1,
                frequency_penalty=0,
                presence_penalty=0
            )
            responses.append(response['choices'][0]['message']['content'].strip())
        except Exception as e:
            responses.append(f"请求失败: {str(e)}")
    return responses

内容的提问来源于stack exchange，提问作者corvusMidnight