Chat

POST /v1/chat

Base URL https://api.unaligned.pw/v1. The body is JSON; Content-Type can be omitted.

Request body

model string required

Model ID, see Models.

prompt string

A single user message. Use either prompt or messages.

messages array

A conversation: [{"role": "user" | "assistant", "content": "…"}]. A user content can also be a list of parts, {"type": "text", "text": "…"} and {"type": "image", "url": "data:image/png;base64,…"} (JPEG, PNG, GIF or WebP, up to 5 MB each, with a model whose vision is true).

system string

Instructions for the model.

max_tokens integer

Maximum tokens to generate. Defaults to what is left of the context window.

temperature number

0 to 2. Lower is more focused. Default 1.

stream boolean

Send the text as it is generated. See Streaming.

Example

curl https://api.unaligned.pw/v1/chat \
  -H "Authorization: Bearer $UNALIGNED_KEY" \
  -d '{
    "model": "gpt-6-luna",
    "system": "You are concise.",
    "messages": [
      {"role": "user", "content": "What is a token?"}
    ],
    "temperature": 0.7
  }'

Response

{
  "id": "req_7f3a…",
  "model": "gpt-6-luna",
  "text": "A token is a piece of text…",
  "finish": "stop",
  "usage": { "input_tokens": 21, "output_tokens": 48 },
  "cost": 0.0000510,
  "balance": 20.4999490
}

text string

The answer.

finish string

stop when the model ended on its own, length when it hit max_tokens.

usage object

Tokens counted for this request.

cost number

What this request cost, in USD.

balance number

Your balance after this request, in USD.