输入灵活性与多模态

使用 OpenAI Responses API

James Chapman

AI Curriculum Manager, DataCamp

输入的灵活性

将提示作为字符串

response = client.responses.create(
    model="gpt-5.4-mini"
    input="How old is the Earth?",
    instructions="Answer concisely."
)

input → 用户输入
instructions → 行为与要求

基于角色的提示

response = client.responses.create(
    model="gpt-5.4-mini"
    input=[
        {"role": "system",
         "content": "Answer concisely."},
        {"role": "user",
         "content": "How old is the Earth?"}
    ]
)

基于角色的消息

response = client.responses.create(
    model="gpt-5.4-mini"
    input=[
        {"role": "system", "content": "Answer concisely."},

        {"role": "user", "content": "How old is the Sun?"},

        {"role": "assistant", "content": "4.603 billion years"},

        {"role": "user", "content": "How old is the Earth?"}

    ]
)

sys_prompt = "You are a helpful Teacher who provides concise, personalized explanations."
latest_response_id = None 

while True:
    user_input = input("You: ").strip()
    if user_input.lower() == "exit":
        break

    response = client.responses.create(
        model="gpt-5.4-mini",
        instructions=sys_prompt,
        input=user_input,
        previous_response_id=latest_response_id
    )

    print(f"\nAssistant: {response.output_text}\n")
    latest_response_id = response.id

messages = [{"role": "system"},
            {"content": "You are a helpful Teacher who provides concise, personalized explanations."}]


while True:
    user_input = input("You: ").strip()
    if user_input.lower() == "exit":
        break

    messages.append({"role": "user", "content": user_input})


    response = client.responses.create(
        model="gpt-5.4-mini",
        input=messages
    )


    print(f"\nAssistant: {response.output_text}\n")
    messages.append({"role": "assistant", "content": response.output_text})

在提示中使用图像！

多模态：通过多种内容模态进行交互

Apple 股价图

¹ 来自 yahoo!finance 的 Apple 公司股价图

在提示中使用图像 URL

response = client.responses.create(
    model="gpt-5.4-mini",
    input=[

        {"role": "user", "content": [
          {"type": "input_text", "text": "Briefly interpret this stock plot."},
          {"type": "input_image", "image_url": "https://...stock_performance.jpg"}]}

    ]
)


print(response.output_text)

在提示中使用图像 URL

- 概要：当日小幅下跌——收于
277.18，较前一日收盘下跌约 0.26%；
盘后小幅走低至 276.94。

Apple 股价图

¹ 来自 yahoo!finance 的 Apple 公司股价图

从本地图像加载

import base64
# 将二进制数据编码为 ASCII 文本字符
with open(image_path, "rb") as f:
    image_base64 = base64.b64encode(f.read()).decode("utf-8")

response = client.responses.create(
    model="gpt-5.4-mini",
    input=[
        {"role": "user", "content": [
          {"type": "input_text", "text": "Briefly interpret this stock plot."},

          {"type": "input_image", "image_url": f"data:image/jpeg;base64,{image_base64}"}]}

    ]
)

让我们来练习！

使用 OpenAI Responses API