Chat Completions — OpenAI-style AI replies

Chat Completion Generation API

What it does

Chat Completions lets you send a list of messages and get a model-generated assistant reply back in the same OpenAI-style format. It is built for conversational applications that need a structured response shape and familiar parameters like model, temperature, top_p, max_tokens, and stream.

Send one or more messages with a role and content, and the service returns a completion object with an id, model, created timestamp, usage counts, and a choices array. Each choice includes the generated message with its role and content, plus a finish_reason so you can handle truncation or completion states in your client logic.

Use Chat Completions when you need to add AI chat to support workflows, internal assistants, content drafting, or task-oriented agents. The response format is predictable, which makes it easier to log, store, and render in applications that already follow chat-style request and response patterns.

Quickstart

Send a single chat-completions request with one user message.

curl -X POST "http://localhost:8080/dosvak/chat-completions" \
  -H "apy-token: $APY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {
        "role": "user",
        "content": "Hello!"
      }
    ]
  }'

What you'll get back

Returns a JSON object with fields like id, model, object, created, usage, and choices. The choices array contains the assistant message, including message.role and message.content.

{
  "id": "chatcmpl-123",
  "model": "Qwen/Qwen3-0.6B",
  "object": "chat.completion",
  "created": 1710000000,
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 24,
    "total_tokens": 36
  },
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello!"
      },
      "finish_reason": "stop"
    }
  ]
}

About this endpoint

What it does

Sends a chat-completions request using an OpenAI-style message array and returns a JSON object containing the generated completion metadata and choices. The request is sent in the body, and the response includes the completion id, model, usage, object, choices, and created fields.

Request Body

Parameter	Type	Mandatory	Description
model	String	No	Model name. Default: `Qwen/Qwen3-0.6B`.
top_p	Number	No	Nucleus sampling value. Must be between `0` and `1`.
stream	Boolean	No	Whether to stream the response. Default: `false`.
messages	Object Array	Yes	Chat messages to send. Must contain at least 1 item. Each item requires `role` and `content`.
messages[].role	ENUM	Yes	Message role. Allowed values: `system`, `user`, `assistant`, `tool`.
messages[].content	String	Yes	Message content.
max_tokens	Integer	No	Maximum number of tokens to generate. Default: `256`. Must be between `1` and `8192`.
temperature	Number	No	Sampling temperature. Default: `0.7`. Must be between `0` and `2`.

Response

Returns a JSON object with id, model, usage, object, choices, and created fields. The usage field is an object containing total_tokens, prompt_tokens, and completion_tokens; choices is an array of objects, each with index, message, and finish_reason.

Parameter	Type	Mandatory	Description
id	String	No	Completion identifier.
model	String	No	Model used for the response.
usage	Object	No	Token usage information. Contains `total_tokens`, `prompt_tokens`, and `completion_tokens`.
usage.total_tokens	Integer	No	Total tokens used.
usage.prompt_tokens	Integer	No	Prompt tokens used.
usage.completion_tokens	Integer	No	Completion tokens used.
object	String	No	Response object type.
choices	Object Array	No	Array of completion choices.
choices[].index	Integer	No	Choice index.
choices[].message	Object	No	Message returned for the choice. Contains `role` and `content`.
choices[].message.role	String	No	Role of the returned message.
choices[].message.content	String	No	Returned message content.
choices[].finish_reason	String	No	Reason the generation stopped.
created	Integer	No	Creation timestamp.

Headers

Name

Type

Description

x-rapidapi-proxy-secretOPTIONAL

string

—

Body

Name

Type

Description

bodyREQUIRED

object

—

Quickstart

What you'll get back

About this endpoint

What it does

Request Body

Response

Headers

Body

Errors any endpoint can return