Xiaomi MiMo Thinking Mode
Toggle thinking mode with the `use_thinking` flag.
Toggle thinking mode with the use_thinking flag. use_thinking=True makes the model emit reasoning_content before its answer; use_thinking=False disables it. Latency and token usage depend on the task and returned output. Leaving it unset (None) uses the model default.
"""
Xiaomi MiMo Thinking Mode
=========================
Toggle thinking mode with the `use_thinking` flag. `use_thinking=True` makes the
model emit `reasoning_content` before its answer; `use_thinking=False` turns it
off for a faster, cheaper response. Leaving it unset (None) uses the model default.
"""
from agno.agent import Agent
from agno.models.xiaomi import MiMo
# ---------------------------------------------------------------------------
# Thinking enabled - returns reasoning_content
# ---------------------------------------------------------------------------
thinking_agent = Agent(
model=MiMo(id="mimo-v2.5-pro", use_thinking=True),
markdown=True,
)
# ---------------------------------------------------------------------------
# Thinking disabled - faster, no reasoning_content
# ---------------------------------------------------------------------------
non_thinking_agent = Agent(
model=MiMo(id="mimo-v2.5-pro", use_thinking=False),
markdown=True,
)
# ---------------------------------------------------------------------------
# Run Agents
# ---------------------------------------------------------------------------
if __name__ == "__main__":
thinking_agent.print_response("Why is the sky blue?", stream=True)
non_thinking_agent.print_response("Why is the sky blue?", 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 openaiExport your Xiaomi MiMo API key
export MIMO_API_KEY="your_mimo_api_key_here"Run the example
Save the code above as thinking_mode.py, then run:
python thinking_mode.pyFull source: cookbook/90_models/xiaomi/thinking_mode.py