Installation¶
Prerequisites¶
Hardware¶
Bit-Axon runs exclusively on Apple Silicon. You need a Mac with an M1, M2, M3, or M4 series chip. Intel-based Macs are not supported.
Operating system¶
macOS 13 (Ventura) or later is required.
Python¶
Python 3.10 or later. If you don't have it yet, the easiest path is Homebrew:
Tip
If you work with multiple Python versions, consider using uv or pyenv to manage your environments. Bit-Axon works with any Python from 3.10 through 3.13.
MLX¶
You don't need to install MLX separately. It's declared as a dependency in the bit-axon package and installs automatically with pip.
Install the package¶
This pulls in all runtime dependencies: mlx, numpy, tokenizers, huggingface_hub, typer, and rich.
Verify the installation¶
If both commands print output without errors, you're good to go.
SwiftUI App (optional)¶
Bit-Axon ships a native macOS chat application built with SwiftUI. To use it, you need:
- Xcode 15 or later
- macOS 14 (Sonoma) or later
Build and run from Xcode. The app provides real-time token streaming, GPU memory monitoring, and drag-and-drop fine-tuning.
Warning
The SwiftUI app requires Xcode and macOS 14+. If you only want the CLI and Python API, you can skip this section entirely.
Development setup¶
If you plan to contribute or work on the codebase, install the package in editable mode with dev dependencies:
The [dev] extra adds pytest, pytest-xdist, ruff, and pre-commit.
Pre-commit hooks¶
Set up the pre-commit hooks to catch lint issues before you push:
Ruff¶
Ruff handles both linting and formatting. Run it manually with:
The project configuration lives in pyproject.toml under [tool.ruff].
Troubleshooting¶
"No module named mlx": MLX requires Apple Silicon. If you're on an Intel Mac, this package won't work. If you are on Apple Silicon, try reinstalling with pip install --force-reinstall mlx.
"Python 3.9 or earlier": Bit-Axon needs Python 3.10+. Check your version with python3 --version and upgrade if needed.
"Command not found: bit-axon": The pip-installed script might not be on your PATH. Try python3 -m bit_axon instead, or make sure your pip bin directory is on PATH.
See also¶
- Quickstart — Download the model and run your first inference
- FAQ — Troubleshooting MLX installation and other common issues