Gemini Interactions - Basic
Run a GeminiInteractions agent through sync, streaming, and async response calls.
Basic example using the Gemini Interactions API.
The four source calls are independent runs. To chain a conversation, add from agno.db.in_memory import InMemoryDb, then configure the agent with db=InMemoryDb() and add_history_to_context=True. This keeps assistant metadata available so later requests can include previous_interaction_id; use durable storage for continuation across restarts.
"""
Gemini Interactions - Basic
============================
Basic example using the Gemini Interactions API.
The Interactions API provides server-side conversation history management,
so only new messages are sent each turn instead of the full history.
"""
import asyncio
from agno.agent import Agent
from agno.models.google import GeminiInteractions
# ---------------------------------------------------------------------------
# Create Agent
# ---------------------------------------------------------------------------
agent = Agent(
model=GeminiInteractions(id="gemini-3.7-flash"),
markdown=True,
)
# ---------------------------------------------------------------------------
# Run Agent
# ---------------------------------------------------------------------------
if __name__ == "__main__":
# --- Sync ---
agent.print_response("Share a 2 sentence horror story")
# --- Sync + Streaming ---
agent.print_response("Share a 2 sentence horror story", stream=True)
# --- Async ---
asyncio.run(agent.aprint_response("Share a 2 sentence horror story"))
# --- Async + Streaming ---
asyncio.run(agent.aprint_response("Share a 2 sentence horror story", stream=True))Run the Example
Set up your virtual environment
uv venv --python 3.12
source .venv/bin/activateInstall dependencies
uv pip install -U agno google-genaiExport your Google API key
export GOOGLE_API_KEY="your_google_api_key_here"Run the example
Save the code above as basic.py, then run:
python basic.pyFull source: cookbook/90_models/google/gemini_interactions/basic.py