Start hereTutorial

Your first request

A copy-paste request, what each field means, and what comes back.

By the end: You can call any text model and read the response and usage figures.

This page takes one request apart, field by field, and shows what comes back.

Request

curl https://www.ary4n73.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTEFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "messages": [
      { "role": "system", "content": "Answer briefly." },
      { "role": "user", "content": "What is an API gateway?" }
    ],
    "max_tokens": 200
  }'

Fields that matter

model — required. Must be an ID from GET /v1/models.
messages — the conversation, oldest first. system sets behaviour.
max_tokens — a ceiling on output. Set it: it bounds a runaway response and its cost.
stream — true to receive server-sent events instead of one JSON body.

Reading the response

Response

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "claude-sonnet-4-6",
  "choices": [
    { "index": 0, "message": { "role": "assistant", "content": "..." }, "finish_reason": "stop" }
  ],
  "usage": { "prompt_tokens": 28, "completion_tokens": 54, "total_tokens": 82 }
}

The usage block is what you are billed on. Input tokens are the prompt, output tokens are the reply. The usage fields reference explains the variations between providers.

Using the official SDKs

OpenAI SDK

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://www.ary4n73.com/v1',
  apiKey: process.env.ROUTEFORGE_API_KEY,
});

const res = await client.chat.completions.create({
  model: 'claude-sonnet-4-6',
  messages: [{ role: 'user', content: 'Hello' }],
});
console.log(res.choices[0].message.content);

The official OpenAI SDK can call Anthropic and Gemini models through this endpoint, because the gateway accepts the OpenAI request shape and returns the OpenAI response shape.