Skip to main content

Installation

RamaLama can be installed on multiple platforms using various methods. Choose the installation method that best fits your environment.

Quick Install​

Universal Install Script (Linux and macOS)​

The easiest way to install RamaLama is using the universal install script:

curl -fsSL https://ramalama.ai/install.sh | bash

This script will automatically detect your system and install RamaLama with the appropriate method.

Platform-Specific Installation​

Fedora​

On Fedora systems, you can install RamaLama directly from the official repositories:

sudo dnf install python3-ramalama

PyPI (All Platforms)​

RamaLama is available on PyPI and can be installed using pip:

pip install ramalama

Optional Components​

MLX Runtime (macOS with Apple Silicon)​

For macOS users with Apple Silicon hardware (M1, M2, M3, or later), you can install the MLX runtime for enhanced performance:

# Using uv (recommended)
uv pip install mlx-lm

# Or using pip
pip install mlx-lm
note

The MLX runtime is specifically designed for Apple Silicon Macs and provides optimized AI model inference. To use MLX, you'll need to run RamaLama with the --nocontainer option.

Verify Installation​

After installation, verify that RamaLama is working correctly:

ramalama version

You should see output similar to:

ramalama version 0.11.1

Next Steps​

Once RamaLama is installed, you can:

  1. Pull your first model: ramalama pull ollama://tinyllama
  2. Run a model: ramalama run ollama://tinyllama
  3. Explore available commands: ramalama --help

For detailed usage instructions, see the Commands section.

Platform-Specific Setup​

After installation, you may need additional platform-specific configuration: