API

Chat Completions

POST /v1/chat/completions takes OpenAI's request body and returns OpenAI's response shape.

POST /v1/chat/completions takes the same body as OpenAI's Chat Completions API and returns the same shape. Every text model on sale answers on this route.

Request body

Parameter	Type		Description
`model`	string	required	A slug from the model list, for example `qwen/qwen3.8-flash`.
`messages`	array	required	The conversation so far, as OpenAI `{ role, content }` objects.
`max_tokens`	integer		Upper bound on output tokens, reasoning included.
`temperature`	number		Sampling temperature.
`tools`	array		Functions the model may call, in OpenAI's `function` tool format.
`tool_choice`	string or object		`auto` lets the model decide. Forcing a call with `required` or a named function depends on the model: `qwen/qwen3.8-flash` refuses both with 400 while it reasons.
`response_format`	object		Ask for JSON output, as in OpenAI's API.
`stream`	boolean		Send the reply as server-sent events. See Streaming.

Only function tools are accepted. Server-side tools that cost extra, such as web search, are refused with 400 unsupported_paid_feature.

Example

Shell

curl https://api.azrouter.dev/v1/chat/completions \
  -H "Authorization: Bearer $AZROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen/qwen3.8-flash",
    "messages": [{"role": "user", "content": "Reply with exactly: Hello from AZRouter"}],
    "max_tokens": 200
  }'

Response (abridged):

JSON

{
  "object": "chat.completion",
  "model": "qwen/qwen3.8-flash",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {"role": "assistant", "content": "Hello from AZRouter"}
    }
  ],
  "usage": {
    "prompt_tokens": 69,
    "completion_tokens": 34,
    "total_tokens": 103,
    "prompt_tokens_details": {"cached_tokens": 0},
    "completion_tokens_details": {"reasoning_tokens": 27}
  }
}

Reasoning models also return the model's thinking in message.reasoning. The response can carry other fields that OpenAI's doesn't; ignore any you don't use.

Tool calls

Describe your functions in tools. When the model wants one, the reply has finish_reason: "tool_calls" and the call in message.tool_calls.

Shell

curl https://api.azrouter.dev/v1/chat/completions \
  -H "Authorization: Bearer $AZROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen/qwen3.8-flash",
    "messages": [{"role": "user", "content": "What is the weather in Bangkok?"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Current weather for a city",
        "parameters": {
          "type": "object",
          "properties": {"city": {"type": "string"}},
          "required": ["city"]
        }
      }
    }],
    "max_tokens": 300
  }'

Response (abridged):

JSON

{
  "finish_reason": "tool_calls",
  "message": {
    "role": "assistant",
    "tool_calls": [
      {
        "type": "function",
        "function": {"name": "get_weather", "arguments": "{\"city\": \"Bangkok\"}"}
      }
    ]
  }
}

Run the function yourself, then send its result back as a {"role": "tool", "tool_call_id": ..., "content": ...} message in the next request.