Chatgpt Prompt Runner

Send one or many prompts to ChatGPT and collect the AI-generated replies in structured JSON. Each prompt runs as an independent request and returns the full response including reply text, model name, message ID, conversation ID, citations, and upstream latency.

What it does

• Bulk prompt execution — run a list of prompts in a single actor run
• Structured output — reply text, model slug, message ID, conversation ID, citations
• Error handling — failed prompts are logged with an error field; other prompts continue
• No browser needed — built on a direct API integration

Input

Field	Type	Description
prompts	array	List of prompts to send to ChatGPT
requestTimeoutSecs	integer	Per-request timeout in seconds (default: 60, max: 180)

Example input

{
  "prompts": [
    "What are the top 5 programming languages in 2025?",
    "Explain the difference between REST and GraphQL.",
    "Write a Python function to reverse a string."
  ],
  "requestTimeoutSecs": 60
}

Output

Each record corresponds to one prompt:

{
  "promptIndex": 0,
  "prompt": "What are the top 5 programming languages in 2025?",
  "reply": "The top 5 programming languages in 2025 are Python, JavaScript, TypeScript, Rust, and Go...",
  "messageId": "49399df9-633a-40ce-81ee-f3ce22567fde",
  "conversationId": "6a0ad6ea-3da0-83ea-b833-f88381dd652c",
  "modelSlug": "gpt-5-3",
  "status": "finished_successfully",
  "citations": [],
  "upstreamLatency": "4.684s",
  "scrapedAt": "2025-05-18T12:00:00+00:00",
  "error": null
}

Usage tips

• Each prompt is a separate independent request — there is no shared conversation context between prompts
• For longer or more detailed prompts, increase requestTimeoutSecs to 90 or 120
• If a prompt fails, its record will have error set and reply as null — the run continues with remaining prompts
• Use promptIndex to match output records back to your original prompt list