Skip to main content
Metaflow is available for Python 3.6+ on macOS and Linux. You can install it using pip or conda.

Requirements

Before installing Metaflow, ensure you have:
  • Python 3.6 or later (supports Python 3.6, 3.7, 3.8, 3.9, 3.10, 3.11, 3.12, and 3.13)
  • pip or conda package manager

Installation Methods

Install Metaflow from PyPI using pip:
pip install metaflow
This will install Metaflow along with its core dependencies:
  • requests - for HTTP operations
  • boto3 - for AWS integrations

Verify Installation

After installation, verify that Metaflow is properly installed by running:
metaflow version
You should see output showing the installed version of Metaflow.

Optional: Type Stubs

If you’re using a type checker like mypy, you can install type stubs for better IDE support:
pip install metaflow[stubs]

Test Your Installation

Create a simple test flow to validate your installation:
hello.py
from metaflow import FlowSpec, step

class HelloFlow(FlowSpec):
    @step
    def start(self):
        print("Metaflow is working!")
        self.next(self.end)
    
    @step
    def end(self):
        print("Flow completed successfully!")

if __name__ == "__main__":
    HelloFlow()
Run the flow:
python hello.py run
If everything is working correctly, you should see output indicating that your flow executed successfully.
For local development, Metaflow works out of the box without any additional configuration. To leverage cloud compute and production orchestrators, you’ll need to configure your infrastructure.

Next Steps

Quickstart

Create your first Metaflow flow in minutes

Tutorial

Learn Metaflow concepts through hands-on examples

Troubleshooting

Python Version Issues

If you encounter Python version compatibility issues, ensure you’re using Python 3.6 or later:
python --version

Permission Errors

If you get permission errors during installation, try installing with the --user flag:
pip install --user metaflow

Getting Help

If you need help with installation, reach out on the Metaflow Slack community.

Build docs developers (and LLMs) love