Chat - Function Calling

Use function calling / tools with chat models

Chat - Function Calling

Enable the model to call external functions by providing a tools array. The model will generate structured JSON input for the function instead of a text response.

For more details, see the OpenAI Function Calling Guide.

Request

Headers

Header	Required	Description
Content-Type	Yes	application/json
Authorization	Yes	Bearer sk-xxx

Request Body

Parameter	Type	Required	Description
model	string	Yes	Model name (e.g. gpt-5.5)
messages	array	Yes	Array of message objects
tools	array	Yes	Array of tool/function definitions
tools[].type	string	Yes	Always function
tools[].function.name	string	Yes	Function name (a-z, A-Z, 0-9, underscores, dashes, max 64)
tools[].function.description	string	No	Function description
tools[].function.parameters	object	Yes	JSON Schema object defining parameters
tools[].function.parameters.type	string	Yes	Usually object
tools[].function.parameters.properties	object	Yes	Parameter definitions
tools[].function.parameters.required	array	No	Required parameter names
tools[].function.strict	boolean	No	Enable strict schema adherence

Example Request

curl https://kiosapi.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "model": "gpt-5.5",
    "messages": [
      {
        "role": "user",
        "content": "What is the weather like in Paris today?"
      }
    ],
    "tools": [
      {
        "type": "function",
        "function": {
          "name": "get_weather",
          "description": "Get current temperature for a given location.",
          "parameters": {
            "type": "object",
            "properties": {
              "location": {
                "type": "string",
                "description": "City and country e.g. Paris, France"
              }
            },
            "required": ["location"],
            "additionalProperties": false
          },
          "strict": true
        }
      }
    ]
  }'

Python Example

import openai
import json

openai.api_key = "sk-xxx"
openai.base_url = "https://kiosapi.com/v1/"

response = openai.chat.completions.create(
    model="gpt-5.5",
    messages=[
        {"role": "user", "content": "What is the weather like in Paris today?"}
    ],
    tools=[
        {
            "type": "function",
            "function": {
                "name": "get_weather",
                "description": "Get current temperature for a given location.",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "City and country e.g. Paris, France"
                        }
                    },
                    "required": ["location"],
                    "additionalProperties": False
                },
                "strict": True
            }
        }
    ]
)

# The model returns a function call
tool_call = response.choices[0].message.tool_calls[0]
args = json.loads(tool_call.function.arguments)
print(args)  # {"location": "Paris, France"}

Response

When the model decides to call a function, the response will contain tool_calls instead of a text content:

{
  "id": "chatcmpl-Ax2bU1RFE8P0Y9uqKcErQNLcx4dDe",
  "object": "chat.completion",
  "created": 1738634724,
  "model": "gpt-5.5-2024-08-06",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": null,
        "tool_calls": [
          {
            "id": "call_FezxjoWuDV1CL3dITVFOjUzK",
            "type": "function",
            "function": {
              "name": "get_weather",
              "arguments": "{\"location\":\"Paris, France\"}"
            }
          }
        ],
        "refusal": null
      },
      "logprobs": null,
      "finish_reason": "tool_calls"
    }
  ],
  "usage": {
    "prompt_tokens": 65,
    "completion_tokens": 16,
    "total_tokens": 81
  },
  "system_fingerprint": "fp_f3927aa00d"
}

The finish_reason will be tool_calls when the model wants to call a function. Execute the function yourself, then send the result back in a new message with role: "tool".