Create chat completion

Creates a model response for a conversation. Send the full message history in each request — the API is stateless, so the model only sees what you include in messages.

POST https://api.tase.app/v1/chat/completions

Authenticate with your API key in the Authorization: Bearer tase_sk_... header. See the Authentication guide for details.

Request body

Required fields

• **model** (string, required) — ID of the model to use. Currently tase-8b. List available IDs with GET /v1/models.
• **messages** (array, required) — The conversation so far, ordered oldest to newest. Each entry is a message object.

Message objects

• **role** (string, required) — One of system, user, assistant, or tool.
• **content** (string, required) — The text of the message. For tool messages, this is the JSON-encoded result your code produced.
• **tool_call_id** (string, required for role tool) — The ID of the tool call this message responds to, taken from the assistant message that requested it.

Optional fields

• **stream** (boolean, optional) — When true, the response is sent incrementally as server-sent events instead of a single JSON body. Defaults to false.
• **temperature** (number, optional) — Sampling temperature between 0 and 2. Lower values make output more deterministic, higher values more varied.
• **max_tokens** (integer, optional) — Upper bound on the number of tokens generated for the completion.
• **tools** (array, optional) — Function definitions the model may call. Each entry declares a function with a name, description, and JSON Schema parameters.
• **tool_choice** (string or object, optional) — Controls tool use: auto lets the model decide, none disables tools, required forces a tool call, and an object of the form {"type": "function", "function": {"name": "..."}} forces one specific function.
• **response_format** (object, optional) — Constrains the output format, for example to request structured JSON output.

Example request

curl https://api.tase.app/v1/chat/completions \
  -H "Authorization: Bearer $TASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "tase-8b",
    "messages": [
      { "role": "system", "content": "You are a concise assistant." },
      { "role": "user", "content": "Summarize why sleep matters in two sentences." }
    ],
    "temperature": 0.7,
    "max_tokens": 200
  }'

Response

• **id** (string) — Unique identifier for this completion.
• **object** (string) — Always chat.completion for non-streaming responses.
• **created** (integer) — Unix timestamp of when the completion was created.
• **model** (string) — The model that produced the response.
• **choices** (array) — The generated completions. Each choice has an index, a message, and a finish_reason.
• **choices[].message** (object) — The assistant message: role, content, and tool_calls when the model requested tools.
• **choices[].finish_reason** (string) — Why generation stopped: stop (natural end), length (hit max_tokens), or tool_calls (the model wants your code to run a tool).
• **usage** (object) — Token accounting: prompt_tokens, completion_tokens, and total_tokens.

Example response

{
  "id": "chatcmpl-9f3a2b1c8d7e6f5a",
  "object": "chat.completion",
  "created": 1784937600,
  "model": "tase-8b",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Sleep is when your body repairs tissue and your brain consolidates memory. Skipping it degrades focus, mood, and long-term health faster than almost any other habit."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 38,
    "total_tokens": 66
  }
}

Tool calls

When you pass tools and the model decides to use one, the assistant message contains a tool_calls array instead of final text, and finish_reason is tool_calls:

{
  "id": "chatcmpl-2c4e6a8b0d1f3e5c",
  "object": "chat.completion",
  "created": 1784937655,
  "model": "tase-8b",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": null,
        "tool_calls": [
          {
            "id": "call_7d1a9c3b5e2f4a68",
            "type": "function",
            "function": {
              "name": "get_weather",
              "arguments": "{\"city\": \"Istanbul\"}"
            }
          }
        ]
      },
      "finish_reason": "tool_calls"
    }
  ],
  "usage": {
    "prompt_tokens": 74,
    "completion_tokens": 21,
    "total_tokens": 95
  }
}

Run the function in your own code, then continue the conversation by appending the assistant message followed by a tool message whose tool_call_id matches the call ID. The model then produces a final answer that incorporates the result.

Streaming

Set "stream": true to receive the completion as server-sent events. Each event is a data: line containing a chunk object; the text arrives incrementally in choices[].delta.content. The stream ends with a literal data: [DONE] line.

data: {"id":"chatcmpl-9f3a2b1c8d7e6f5a","object":"chat.completion.chunk","created":1784937600,"model":"tase-8b","choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":null}]}

data: {"id":"chatcmpl-9f3a2b1c8d7e6f5a","object":"chat.completion.chunk","created":1784937600,"model":"tase-8b","choices":[{"index":0,"delta":{"content":"Sleep is when"},"finish_reason":null}]}

data: {"id":"chatcmpl-9f3a2b1c8d7e6f5a","object":"chat.completion.chunk","created":1784937600,"model":"tase-8b","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]