This guide covers setting up DeepSeek R1 Local on macOS, including both CPU and GPU (Metal) support.
Note: I have not tested this on macOS. I've only verifed on Linux(Ubuntu) and Windows.
- macOS 10.15 or later
- Apple Silicon (M1/M2/M3) or Intel processor
-
Install Docker Desktop for macOS:
- Download from Docker's official website
- For Apple Silicon Macs, ensure you download the Apple Silicon version
- Follow the installer prompts
- Start Docker Desktop and wait for it to complete initialization
-
Install make (if not already installed):
- Using Homebrew (recommended):
brew install make
- Or using Xcode Command Line Tools:
xcode-select --install
- Using Homebrew (recommended):
-
Start the service:
For CPU-only:
docker compose up -d
For GPU (Metal) support:
docker compose -f docker-compose.yml -f docker-compose.mac.yml up -d
- Native arm64 support provides excellent performance
- GPU acceleration via Metal is available out of the box
- Memory usage is typically lower than on Intel Macs
- CPU-only operation
- May experience higher memory usage
- Consider resource limits in Docker Desktop settings
-
If Docker fails to start:
- Check Docker Desktop is running
- Ensure you have enough memory allocated in Docker Desktop settings
- Try restarting Docker Desktop
-
For performance issues:
- In Docker Desktop, adjust memory and CPU allocation:
- Go to Preferences → Resources
- Recommended minimum: 8GB RAM, 4 CPUs
- Check activity monitor for resource usage
- In Docker Desktop, adjust memory and CPU allocation:
-
For Metal acceleration issues (Apple Silicon):
- Ensure macOS is updated to the latest version
- Check Docker Desktop settings for GPU support
- Verify Metal support:
docker run --rm -it --platform linux/arm64 ubuntu:latest apt-get update && apt-get install -y clinfo && clinfo