OpenAI API payload → typed code

To type an OpenAI API payload, paste a chat.completion response into PayloadIQ and it generates TypeScript interfaces and a Zod schema — in your browser, nothing uploaded. The response has a consistent shape (choices[].message, usage, finish_reason) that is worth capturing as types so your code stops indexing into any.

Example chat completion response

A typical non-streaming response from the Chat Completions endpoint:

{
  "id": "chatcmpl-AbC123dEf456",
  "object": "chat.completion",
  "created": 1727200000,
  "model": "gpt-4o-mini-2024-07-18",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help you today?",
        "refusal": null
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 9,
    "completion_tokens": 12,
    "total_tokens": 21
  },
  "system_fingerprint": "fp_0123456789"
}

What you get back

PayloadIQ infers a ChatCompletion interface with a typed choices array, the nested message object, and a usage block of token counts — plus a Zod schema you can use to validate responses before trusting choices[0].message.content. It is a fast way to bootstrap a typed wrapper around the API without hand-writing the model.

Browser-local by design

The payload is parsed and transformed on your device — useful when a response includes proprietary prompt content or tool-call arguments you would rather not paste into a random online generator.