Chatgpt Master Guide

| Parameter | Type | Required | Description | |---|---:|---:|---| | model | string | Yes | Model identifier (example: "gpt-4.1-mini"). | | input | string \| array | Yes | Prompt content. Can be a single string or an array of message items. | | max_output_tokens | integer | No | Upper bound on generated tokens (default is model-dependent). | | temperature | number | No | Sampling temperature (commonly 0 - 2). Lower is more deterministic. | | top_p | number | No | Nucleus sampling probability (common range 0 - 1). | | response_format | object | No | Controls response shape. Example: { "type": "json_object" }. | | metadata | object | No | Key/value tags echoed back in usage for tracking. | | stream | boolean | No | If true, returns incremental events (SSE). Default false. |

Authentication (header) | Header | Type | Required | Description | |---|---:|---:|---| | Authorization | string | Yes | Must be exactly Bearer. | | Content-Type | string | Yes | Use application/json. |

Code Example

`bash

1) Export your API key as an environment variable (Key-Chain Setup Method)

Keep the key out of source code and logs. export OPENAI_API_KEY="put_your_key_here"

2) Call the Responses API with the Bearer token header. curl https://api.openai.com/v1/responses \ -H "Authorization: Bearer $OPENAI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "gpt-4.1-mini", "input": "Return a JSON object with keys: status and note.", "response_format": { "type": "json_object" }, "max_output_tokens": 120, "temperature": 0 }' `

`python

Python example using the same authentication pattern. import os import requests

api_key = os.environ["OPENAI_API_KEY"]

Key-Chain Setup Method: read from env

url = "https://api.openai.com/v1/responses" headers = { "Authorization": f"Bearer {api_key}", "Content-Type": "application/json", }

payload = { "model": "gpt-4.1-mini", "input": "Return a JSON object with keys: status and note.", "response_format": {"type": "json_object"}, "max_output_tokens": 120, "temperature": 0, }

resp = requests.post(url, headers=headers, json=payload, timeout=30) data = resp.json() print(data) `

Response Format

json { "id": "resp_12345", "object": "response", "created_at": 1720000000, "model": "gpt-4.1-mini", "output": [ { "id": "out_1", "type": "message", "role": "assistant", "content": [ { "type": "output_text", "text": "{\"status\":\"ok\",\"note\":\"...\"}" } ] } ], "usage": { "input_tokens": 12, "output_tokens": 34, "total_tokens": 46 } }

Field notes - output: Array of generated outputs; content holds the actual text (or other content types). - usage: Token counts for billing and debugging. - response_format: json_object: Typically results in JSON text inside output_text.text.

Notes & Best Practices

• Key-Chain Setup Method: Store the API key in an environment variable (OPENAI_API_KEY) and read it at runtime; avoid committing keys to Git and avoid printing headers in logs. - Validate permissions: 401 usually means the key is missing/invalid; 403 often means the key is present but restricted (organization/project permissions). - Handle rate limits: 429 indicates throttling; implement exponential backoff and retry only idempotent request patterns. - Use exact header format: Authorization: Bearer must match exactly; extra spaces or missing Bearer commonly cause 401.

Authentication and request shaping determine whether your inputs reach the model; next, you’ll focus on how to structure input and control generation so outputs stay consistent.

End of chapter one. 4 more chapters in the full book.