Completions

POST /v1/completions

Use this endpoint for legacy text completion clients and code fill-in-the-middle flows.

Request

{
  "model": "qwen3.5-coder-32b",
  "prompt": "def fibonacci(n):",
  "max_tokens": 128,
  "temperature": 0.2
}

Fill in the middle

{
  "model": "qwen3.5-coder-32b",
  "prompt": "def add(a, b):\n    ",
  "suffix": "\nprint(add(2, 3))",
  "max_tokens": 64
}

Response

{
  "id": "cmpl_...",
  "object": "text_completion",
  "choices": [
    {
      "text": "return a + b",
      "index": 0,
      "finish_reason": "stop"
    }
  ]
}