Tool calling

Let the model call functions you define — OpenAI-compatible tools.

Pass a tools array; the model returns tool_calls when it decides to use one. tool_choice supports "auto" (default), "none", or a named function.

curl https://api.sara-ai.kz/v1/chat/completions \
  -H "Authorization: Bearer $SARA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma-4-31b",
    "messages": [{"role": "user", "content": "Weather in Astana? Use the tool."}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Get current weather for a city",
        "parameters": {
          "type": "object",
          "properties": { "city": { "type": "string" } },
          "required": ["city"]
        }
      }
    }]
  }'
# response: choices[0].message.tool_calls[0].function

Small models call tools less reliably than large ones — validate arguments server-side and design for retries.