POST /v1/chat/completions

OpenAI-compatible chat. The response arrives in the same HTTP request.

Endpoint


POST https://llm.dat.ai/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

Request body

JSON object. For the full shape — including messages — see the example below.

Field	Required	Description
`model`	Yes	Model name, e.g. `qwen3:1.7b`
`messages`	Yes	Objects with `role` and `content` (roles: `user`, `assistant`, `system`)
`stream`	No	Default `false`. Set `true` for SSE streaming
`max_tokens`, `max_completion_tokens`	No	Max tokens to generate
`temperature`, `top_p`, `stop`, `seed`	No	Standard OpenAI sampling options
`tools`, `tool_choice`	No	OpenAI function calling — see Function calling below
`datai`	No	dat.ai built-in node tools — see Inference overview

Optional header: X-Request-ID (correlation ID; server generates a UUID if omitted).

Function calling

Use tools and tool_choice for OpenAI function calling : declare functions the model can call, and optionally control selection with tool_choice (auto, none, required, or a specific function).

tools — array of { "type": "function", "function": { "name", "description", "parameters" } } where parameters is a JSON Schema object
tool_choice — whether/how the model picks a tool

These fields are forwarded to the node unchanged. They are not the same as datai.tools (net, fs), which turn on dat.ai built-in node capabilities.

Response

Non-streaming (200): standard OpenAI chat.completion:


{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "qwen3:1.7b",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello!" },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 10,
    "completion_tokens": 5,
    "total_tokens": 15
  }
}

usage is present only when the node reports token counts.

Streaming: Content-Type: text/event-stream. Chunks follow the OpenAI streaming format; the stream ends with data: [DONE].

Errors

OpenAI-style JSON — see Authentication. Common cases:

Status	Cause
400	Invalid JSON, missing `model`/`messages`, `datai.tools` + `stream: true`
401	Missing or invalid API key
402	Insufficient credits
504	Request timed out (max wait 180 seconds)

Example


curl -sS -X POST "https://llm.dat.ai/v1/chat/completions" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3:1.7b",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": false
  }'