Extended Thinking

Enable extended thinking for a Claude agent through Azure AI Foundry.

Manual thinking={"type": "enabled", "budget_tokens": ...} still works on Sonnet 4.6, but is deprecated there. Prefer adaptive thinking for new configurations, and check model-specific support before changing the deployment ID.

Code

from agno.agent import Agent
from agno.models.azure.claude import Claude

agent = Agent(
    model=Claude(
        id="claude-sonnet-4-6",
        max_tokens=2048,
        thinking={"type": "enabled", "budget_tokens": 1024},
    ),
    markdown=True,
)

agent.print_response("Share a very scary 2 sentence horror story")

Usage

Set up your virtual environment

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

Set your environment variables

export ANTHROPIC_FOUNDRY_API_KEY=xxx
export ANTHROPIC_FOUNDRY_RESOURCE=xxx

Install dependencies

uv pip install -U anthropic agno

Run Agent

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

python thinking.py