> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agno.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Moonshot Reasoning Effort

> Kimi K3 always reasons before answering.

Kimi K3 always reasons before answering. How much it thinks is controlled by the top-level `reasoning_effort` parameter, which accepts three levels - "low", "high" and "max" - and defaults to "max" when omitted.

```python reasoning_effort.py theme={null}
"""
Moonshot Reasoning Effort
=========================

Kimi K3 always reasons before answering. How much it thinks is controlled by the
top-level `reasoning_effort` parameter, which accepts three levels - "low", "high" and
"max" - and defaults to "max" when omitted.

"max" is a strong default - K3 can spend a minute or more thinking before answering a
prompt that does not need it. "high" is the middle ground, and "low" is several times
faster at the cost of shallower reasoning. Use "max" for genuinely hard problems, "high"
for everyday work that still benefits from some thought, and "low" where latency matters
more than depth.

The reasoning is returned as reasoning_content, which `show_full_reasoning=True`
renders alongside the answer.
"""

from agno.agent import Agent
from agno.models.moonshot import MoonShot

# ---------------------------------------------------------------------------
# Deep reasoning - the default, worth it for a hard problem
# ---------------------------------------------------------------------------

deep_agent = Agent(
    model=MoonShot(id="kimi-k3", reasoning_effort="max"),
    markdown=True,
)

task = (
    "A farmer needs to cross a river with a fox, a chicken and a sack of grain. "
    "The boat only fits the farmer and one item. The fox cannot be left alone with "
    "the chicken, and the chicken cannot be left alone with the grain. "
    "Provide a step-by-step solution."
)

# ---------------------------------------------------------------------------
# Balanced reasoning - the middle ground, some thought without the full cost
# ---------------------------------------------------------------------------

balanced_agent = Agent(
    model=MoonShot(id="kimi-k3", reasoning_effort="high"),
    markdown=True,
)

# ---------------------------------------------------------------------------
# Low reasoning - much faster, for prompts that do not need deep thought
# ---------------------------------------------------------------------------

fast_agent = Agent(
    model=MoonShot(id="kimi-k3", reasoning_effort="low"),
    markdown=True,
)

# ---------------------------------------------------------------------------
# Run Agents
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    deep_agent.print_response(task, stream=True, show_full_reasoning=True)

    balanced_agent.print_response(task, stream=True, show_full_reasoning=True)

    fast_agent.print_response(task, stream=True, show_full_reasoning=True)
```

## Run the Example

<Steps>
  <Snippet file="create-venv-step.mdx" />

  <Step title="Install dependencies">
    ```bash theme={null}
    uv pip install -U agno openai
    ```
  </Step>

  <Step title="Export your Moonshot API key">
    <CodeGroup>
      ```bash Mac/Linux theme={null}
      export MOONSHOT_API_KEY="your_moonshot_api_key_here"
      ```

      ```bash Windows theme={null}
      $Env:MOONSHOT_API_KEY="your_moonshot_api_key_here"
      ```
    </CodeGroup>
  </Step>

  <Step title="Run the example">
    Save the code above as `reasoning_effort.py`, then run:

    ```bash theme={null}
    python reasoning_effort.py
    ```
  </Step>
</Steps>

Full source: [cookbook/90\_models/moonshot/reasoning\_effort.py](https://github.com/agno-agi/agno/blob/v3.0.4/cookbook/90_models/moonshot/reasoning_effort.py)