OpenAI API Beginner: Your First GPT-5.6 Call Explained

The OpenAI API is the most direct way to integrate GPT models into your own product. This guide walks you from zero to your first call.

1. Create an OpenAI Account

1. Visit platform.openai.com
2. Sign up with email or Google
3. Verify your phone number (note: mainland China numbers may face restrictions)
4. Add a payment method (credit card)
5. Top up your Billing page (minimum $5)

2. Get an API Key

Path: API Keys → Create new secret key → name it (e.g. my-app-dev) → create.

⚠️ The key is shown only once. Copy it immediately and store it somewhere safe.

3. Set Up Environment Variables

Never hardcode the API key. Use environment variables:

macOS / Linux:

export OPENAI_API_KEY="sk-..."

Windows PowerShell:

$env:OPENAI_API_KEY="sk-..."

.env file (project root):

OPENAI_API_KEY=sk-...

Add .env to .gitignore.

4. Python — Your First Call

Install the SDK:

pip install openai

Sample code (Responses API):

from openai import OpenAI

client = OpenAI()  # auto-reads OPENAI_API_KEY from env

response = client.responses.create(
    model="gpt-5.6",
    input=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain what ChatGPT is in one sentence."},
    ],
    reasoning={"effort": "medium"},
    max_output_tokens=200,
)

print(response.output_text)
print(f"Tokens used: {response.usage.total_tokens}")

5. Node.js — Your First Call

Install the SDK:

npm install openai

Sample code:

import OpenAI from 'openai';

const client = new OpenAI(); // auto-reads process.env.OPENAI_API_KEY

const response = await client.responses.create({
  model: 'gpt-5.6',
  input: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'Explain what ChatGPT is in one sentence.' },
  ],
  reasoning: { effort: 'medium' },
  max_output_tokens: 200,
});

console.log(response.output_text);
console.log('Tokens used:', response.usage?.total_tokens);

6. Multi-turn Conversation

Pass the whole conversation as the input array:

response = client.responses.create(
    model="gpt-5.6",
    input=[
        {"role": "system", "content": "You are a Python mentor"},
        {"role": "user", "content": "What is a decorator?"},
        {"role": "assistant", "content": "A decorator is..."},
        {"role": "user", "content": "Can you give me an example?"},
    ],
)

Or just pass the previous response ID and let the API stitch the history:

response = client.responses.create(
    model="gpt-5.6",
    input="What is a decorator?",
    previous_response_id=prev_id,
)

7. Streaming Response (SSE)

Show tokens as they arrive for a smoother UX:

stream = client.responses.create(
    model="gpt-5.6",
    input=messages,
    stream=True,
)

for event in stream:
    if event.type == "response.output_text.delta":
        print(event.delta, end="")

8. Tools and Function Calling

The Responses API mounts tools via the tools field: Web search, file search, computer use, Code Interpreter, and custom functions.

response = client.responses.create(
    model="gpt-5.6",
    input="What's the weather in Beijing today?",
    tools=[{"type": "web_search"}],
)

9. Error Troubleshooting

• 401 invalid_api_key — check the API Key
• 429 rate_limit_exceeded — back off and add retries, or switch to Luna
• 402 insufficient_quota — top up your account
• 400 invalid_request_error — common cause: passing messages to the Responses API (use input instead)
• timeout — network problems; retry or change region

10. Cost Optimisation

• Use Luna for high-throughput batch jobs
• Enable Prompt Caching for repeated prefixes
• Compress system prompts
• Cap output with max_output_tokens
• Pre-summarise long inputs
• Set reasoning.effort = "none" for trivial tasks

11. Migrating from Chat Completions

If you have existing Chat Completions code, the move to Responses API is usually a few lines:

- response = client.chat.completions.create(
-     model="gpt-4o",
-     messages=[{"role": "user", "content": "..."}],
+ response = client.responses.create(
+     model="gpt-5.6",
+     input=[{"role": "user", "content": "..."}],
  )
- text = response.choices[0].message.content
+ text = response.output_text

Chat Completions still works but is no longer gaining new features. New projects should use the Responses API directly.

12. Next Steps

Once you are comfortable here, continue with:

• The complete guide to GPT models: GPT-5.6 Sol, Terra, Luna
• Prompt Engineering Core Patterns: 8 Templates That 2x GPT Output
• Realtime Voice guide: gpt-realtime and Voice Mode