Gemini Cookbook: Official Quickstarts · Lesson 2

Function Calling in the Gemini API

Pass functions as tools, enable automatic_function_calling in ChatSession, inspect FunctionCall/FunctionResponse history, and implement a manual calling loop.

20 min read3 questions in quizReady prompt includedIn progress

Practical exercise

What to do after this lesson

Build a chatbot with three tools: get_weather(city), convert_currency(amount, from_currency, to_currency), and calculate_tax(income, rate). Test a multi-tool scenario with a single user message.

Ready-to-use prompt

Template for this lesson

Copy and adapt to your context. Text in angle brackets should be replaced.

from google import genai

client = genai.Client(api_key=GOOGLE_API_KEY)

def get_stock_price(ticker: str) -> dict:
    """Get the current stock price for a given ticker symbol."""
    prices = {"GOOGL": 175.0, "AAPL": 190.5, "MSFT": 415.2}
    return {"ticker": ticker, "price": prices.get(ticker, 0), "currency": "USD"}

chat = client.chats.create(
    model="gemini-2.5-flash",
    config={"tools": [get_stock_price]}
)

response = chat.send_message("What is the current price of Apple and Google stocks?")
print(response.text)

Common mistakes

What people get wrong

Using unsupported parameter types (only int, float, bool, str, list, dict are allowed); missing docstrings (the model won't understand the function's purpose); forgetting to send FunctionResponse in manual mode.

Defining Tool Functions

def enable_lights(): """Turn on the lighting system.""" print("LIGHTBOT: Lights enabled.") def set_light_color(rgb_hex: str): """Set the light color. Lights must be enabled for this to work.""" print(f"LIGHTBOT: Lights set to {rgb_hex}.") def stop_lights(): """Stop flashing lights.""" print("LIGHTBOT: Lights turned off.") light_controls = [enable_lights, set_light_color, stop_lights]

Chat with Automatic Function Calling

chat = client.chats.create( model=MODEL_ID, config={ "tools": light_controls, "system_instruction": "You are a lighting system bot.", # automatic_function_calling is enabled by default } ) response = chat.send_message("It's awful dark in here...") print(response.text)

def multiply(a: float, b: float): """returns a * b.""" return a * b chat = client.chats.create( model=MODEL_ID, config={"tools": [multiply]} ) response = chat.send_message( "I have 57 cats, each owns 44 mittens, how many mittens is that?" ) print(response.text)

Inspecting Chat History

for content in chat.get_history(): print(content.role) for part in content.parts: if part.text: print("Text:", part.text) if part.function_call: print("FunctionCall:", part.function_call) if part.function_response: print("FunctionResponse:", part.function_response)

response = client.models.generate_content( model=MODEL_ID, contents="Which theaters in Mountain View show Barbie?", config={"tools": theater_functions, "automatic_function_calling": {"disable": True}} ) # Handle FunctionCall manually and send FunctionResponse back

Сообщить об ошибке

Function Calling in the Gemini API

Квиз — 3 вопроса

What Is Function Calling

Defining Tool Functions

Chat with Automatic Function Calling

Math Tools Example

Inspecting Chat History

Manual Mode (generate_content)

Auto Schema Generation