Basic Agent

Run a Cloudflare Workers AI model in sync, async, and streaming modes.

Code

import asyncio

from agno.agent import Agent
from agno.models.cloudflare import Cloudflare

agent = Agent(
    model=Cloudflare("@cf/meta/llama-3.3-70b-instruct-fp8-fast"),
    markdown=True,
)

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))

Usage

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Set your environment variables

export CLOUDFLARE_API_TOKEN=xxx
export CLOUDFLARE_ACCOUNT_ID=xxx
export CLOUDFLARE_AI_GATEWAY_ID=xxx  # optional, defaults to "default"

Install dependencies

uv pip install -U openai agno

Run Agent

Save the code above as basic.py, then run:

python basic.py