Apple Silicon Mac installation with MLX

Use this guide for an M-series Mac. The published macOS binaries target Apple Silicon; this is not an Intel Mac installation guide. MLX is the native Apple Silicon LLM runtime, not a separate operating system.

1. Open Terminal

Run under your normal macOS account. Keep the machine awake and allow enough disk space for the selected model. Do not run the entire installer with sudo.

2. Install the commands

curl -fL --progress-bar --connect-timeout 20 https://github.com/mundusx/releases/releases/download/opengpu-prod/install.sh | bash

The script downloads the OpenGPU CLI and node agent, and displays download/checksum progress. The default is install-only; it prints the next commands rather than immediately starting contribution.

If opengpu is not found, open a new Terminal window. For the current shell:

export PATH="$HOME/.local/bin:$PATH"

3. Choose your contribution and model

opengpu install

Follow the shared wizard guide. First choose MundusX-managed runtime, then select workloads and an offered MLX-compatible model to use the native MLX path. The explicit command is opengpu install --connection managed.

To reuse Ollama or LM Studio instead, start that engine first. Load an LLM in LM Studio and enable its API server, then run (using your actual endpoint):

opengpu install --connection direct --cluster-url http://127.0.0.1:1234

Select the exact model and wait for its verification request. This uses the external engine and skips MundusX MLX/model provisioning. See Inference connections for details. PAIR is currently validation-only and cannot start contribution.

For managed MLX, setup prepares an isolated Python virtual environment and dependencies and downloads the selected model. A Python virtual environment separates packages; it is not an OS security sandbox.

Does this require a .pkg? The MundusX shell path does not require the MundusX .pkg. If Python is already available, setup uses it. If Python is missing, the current MLX bootstrap downloads and installs an official Python package; macOS may request administrator authentication for that dependency. Enter any OS password only in the local OS/terminal prompt, never in Chat.

4. Start and check readiness

opengpu --version
opengpu start --background
opengpu status
opengpu doctor
opengpu model list

This release reports opengpu 0.2.23. Complete onboarding when prompted. Check for readyForJobs: yes, a connected node, and the intended active model. MLX attempts to launch a persistent server and runs a small inference warmup. Startup/model loading can take time. If the persistent path fails, fallback execution can be used; do not assume every successful connection is a warm persistent MLX server. Inspect diagnostics.

5. Job limits and stopping

The Mac memory ceiling is 1 concurrent job below 32 GiB RAM, 2 from 32 GiB up to below 128 GiB, and 4 at 128 GiB or above. Actual model capacity can be lower.

opengpu pause
opengpu resume
opengpu exit

Background execution does not itself promise automatic startup after reboot. For direct connections, start the external engine and load its selected model before opengpu start --background. Managed MLX reloads and warms its saved active model on startup. Configuration and models remain saved. The separate mundusx connect Chat workflow is not required here.

For images/video, see media setup. On Mac, ComfyUI uses PyTorch MPS/Metal, not MLX. Native setup is implemented, but the bundled models must pass real verification on the specific Mac before media readiness is claimed.

Installation progress

The initial script download shows curl progress. The installer then prints its current step and download progress. Slow download, GPU checks, and Docker steps print an elapsed-time update every 10 seconds; elapsed time is not a percentage or an estimate of remaining time. A step failure prints its exit status. Do not launch a second installer while one is still running.