Google Gen AI Python SDK

Use client.chats.create to initialize a chat session. You can then use send_message for synchronous non-streaming responses, send_message_stream for synchronous streaming, or the asynchronous equivalents via client.aio.chats.create for send_message and send_message_stream.

# Synchronous Non-Streaming
chat = client.chats.create(model='gemini-3.5-flash')
response = chat.send_message('tell me a story')
print(response.text)

# Synchronous Streaming
chat = client.chats.create(model='gemini-3.5-flash')
for chunk in chat.send_message_stream('tell me a story'):
    print(chunk.text)

# Asynchronous Non-Streaming
chat = client.aio.chats.create(model='gemini-3.5-flash')
response = await chat.send_message('tell me a story')
print(response.text)

# Asynchronous Streaming
chat = client.aio.chats.create(model='gemini-3.5-flash')
async for chunk in await chat.send_message_stream('tell me a story'):
    print(chunk.text)

To bypass automatic calling, manually declare a function using types.FunctionDeclaration, pass it via types.Tool, and handle the response.function_calls by executing the function and passing a types.Part.from_function_response back to the model.

from google.genai import types

function = types.FunctionDeclaration(
    name='get_current_weather',
    description='Get the current weather in a given location',
    parameters_json_schema={
        'type': 'object',
        'properties': {
            'location': {
                'type': 'string',
                'description': 'The city and state, e.g. San Francisco, CA',
            }
        },
        'required': ['location'],
    },
)

tool = types.Tool(function_declarations=[function])

response = client.models.generate_content(
    model='gemini-3.5-flash',
    contents='What is the weather like in Boston?',
    config=types.GenerateContentConfig(tools=[tool]),
)

# Handle the function call manually
function_call_part = response.function_calls[0]
# ... execute function ...
function_response_part = types.Part.from_function_response(
    name=function_call_part.name,
    response={'result': 'sunny'},
)
function_response_content = types.Content(role='tool', parts=[function_response_part])

# Send response back to model
response = client.models.generate_content(
    model='gemini-3.5-flash',
    contents=[
        user_prompt_content, # original user prompt
        response.candidates[0].content, # the model's function call
        function_response_content, # your function result
    ],
    config=types.GenerateContentConfig(tools=[tool]),
)

The Interactions API provides a unified interface for Gemini models and agents, supporting stateful conversations, multimodal input, and tool orchestration.

• Basic Interaction: Use client.interactions.create with a model and input.
• Stateful Conversation: Pass previous_interaction_id to client.interactions.create to continue a conversation.
• Agents (Deep Research): Use specialized agents (e.g., deep-research-pro-preview-12-2025) by passing the agent name. For long-running tasks, set background=True and poll client.interactions.get(id=...) until the status is completed.

# Stateful Conversation
interaction1 = client.interactions.create(model='gemini-3.5-flash', input='Hi, my name is Amir.')
interaction2 = client.interactions.create(
    model='gemini-3.5-flash', 
    input='What is my name?', 
    previous_interaction_id=interaction1.id
)
print(interaction2.outputs[-1].text)