LLM API Request Generator

Generated a OpenAI gpt-4o-mini request; nothing was sent.

How it works

• Resolved the provider endpoint and the placeholder auth header.
• Serialized the model, messages and parameters into the provider's request body.
• Rendered the request as code and attached the disclosure block.

• If you run this request, the model name, messages and parameters are sent from your machine straight to the provider endpoint shown above. This site is not involved and sends nothing.
• Use your own provider API key in place of YOUR_API_KEY. This site never asks for, stores, logs or transmits your key; keep it in a secret store and never paste it somewhere untrusted.

Methodology

1. Resolve a fixed provider endpoint from a closed provider set; the endpoint is never taken from user input.
2. Serialize the model, messages and parameters into the provider's request body with a placeholder credential (`YOUR_API_KEY`), never a real key.
3. Render the request as cURL, Python or raw JSON and attach the mandatory disclosure of provider, model, cost basis, transmission, retention and key handling.

Usage

1. Pick a provider, enter the model id and the user message (and an optional system message).
2. Choose the output language: cURL, Python or JSON.
3. Read the disclosure block, then copy the request and replace YOUR_API_KEY with your own key.

Examples

These are examples, not current values: they use fixed inputs, while the interactive tool computes live results in your browser.

An OpenAI chat request

{
  "provider": "openai",
  "model": "gpt-4o-mini",
  "system": "You are concise.",
  "user": "Summarize RFC 8259 in one sentence.",
  "language": "curl"
}

An Anthropic message request

{
  "provider": "anthropic",
  "model": "claude-sonnet-5.5",
  "user": "List three JSON edge cases.",
  "maxTokens": 256,
  "language": "python"
}

Limitations

Client-direct by design

A generated request is meant to be run by you, against the provider, from your own environment. This page never proxies a model call: it cannot, because it is a static page with no server. That is also why it can promise never to see your key.

Read the disclosure before you run it

The disclosure block names the provider, the model, how cost is billed, what would be transmitted, how long the provider keeps it and how the key is handled. Read it before sending production data.