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

# Installation

> Install QitOS from PyPI or source, with optional extras for models and benchmarks.

<Warning>
  QitOS is currently **Alpha**. The core `AgentModule + Engine` kernel, tracing, and canonical examples are stable. Higher-level APIs may change between releases. Pin your version in production.
</Warning>

## Requirements

QitOS requires **Python 3.10 or later**. It is tested on Python 3.10, 3.11, and 3.12.

<Note>
  Need a remote model API before running LLM-powered examples? Start with [Prerequisites](/prerequisites) for provider suggestions, API key setup, and compatibility checks.
</Note>

## Standard install

Install the core framework from PyPI:

```bash theme={null}
pip install qitos
```

The core install includes the `AgentModule + Engine` kernel, the `qita` observability tools (QitOS's built-in trace viewer), and the `kit` toolkit modules (pre-built parsers, memory adapters, toolsets, critics, and planners). It does not include the OpenAI client or benchmark dataset loaders.

<Tip>
  If you want to run the minimal coding-agent quickstart immediately, skip straight to `pip install "qitos[models]"`.
</Tip>

## Optional extras

Use extras to add support for models, benchmarks, or everything at once.

<CodeGroup>
  ```bash OpenAI support theme={null}
  pip install "qitos[models]"
  ```

  ```bash Benchmark adapters theme={null}
  pip install "qitos[benchmarks]"
  ```

  ```bash Everything theme={null}
  pip install "qitos[all]"
  ```
</CodeGroup>

| Extra        | What it adds                                                                          |
| ------------ | ------------------------------------------------------------------------------------- |
| `models`     | `openai>=1.0.0` — required to use OpenAI-compatible model endpoints                   |
| `benchmarks` | `datasets` and `huggingface_hub` — required for GAIA, Tau-Bench, and CyBench adapters |
| `all`        | All of the above                                                                      |

<Tip>
  If you're just getting started, `pip install "qitos[models]"` is the right choice. It gives you everything you need to run the minimal coding-agent quickstart.
</Tip>

## Source install

Clone the repository and install in editable mode with all development dependencies:

```bash theme={null}
git clone https://github.com/Qitor/qitos.git
cd qitos
pip install -e ".[dev,models,benchmarks]"
```

The `dev` extra adds `pytest`, `black`, `flake8`, `mypy`, and the build tools used by the QitOS team.

## Verify your installation

Confirm the package is installed and check the version:

```bash theme={null}
python -c "import qitos; print(qitos.__version__)"
```

You should see output like:

```
0.3.0
```

You can also verify the minimal coding agent and both CLIs are on your path:

```bash theme={null}
qit demo minimal
qit --help
qita --help
```

## Next steps

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
    Run your first agent in under 2 minutes
  </Card>

  <Card title="Core concepts" icon="book" href="/concepts/agent-module">
    Understand AgentModule, Engine, State, and Tools
  </Card>
</CardGroup>