Use cases

One remote Mac, three ways to use it

Run Xcode development, automated builds, or model inference on a dedicated physical Mac. Use the graphical desktop over screen sharing or VNC, and connect over SSH when you need the command line. Compute resources are not shared with other tenants.

Choose from two M4 configurations and daily, weekly, monthly, or quarterly plans. Check the console for real-time node availability.

One machine, scheduled by taskDedicated physical node · GUI + SSH
01
Xcode workspaceEdit · Simulator · Test distribution
Desktop
02
Automated build queueSelf-hosted runner · Signing · Archiving
CI
03
Local inference sessionMLX / llama.cpp · Output monitoring
Inference

Schedule access by task and user; simultaneous control of the same desktop is not a collaboration workflow.

For iOS and macOS developers

Run Xcode remotely. Keep your workflow in your hands.

Open the macOS desktop over screen sharing or VNC to edit projects, run Xcode, and review build logs remotely. Simulators and compile jobs use the physical Mac’s own CPU, GPU, and unified memory, leaving your local computer’s resources free.

When it’s time to deliver a build to testers, archive, sign, and distribute it through TestFlight on the remote Mac. Manage your project repository, signing assets, and export settings separately. Before your first distribution, verify certificates, provisioning profiles, and build settings with an internal test build.

Limits of device debugging

A cloud Mac can’t connect directly to the iPhone beside you. Keep simulator runs, builds, and distribution on the remote Mac; use a local device and development environment when you need a USB connection, device logs, or hands-on testing.

A development session

  1. 01
    Connect to the desktop

    Choose a nearby node, check the display resolution, and open your project workspace.

  2. 02
    Build and verify

    Run the target simulator first, then check memory use and build output.

  3. 03
    Deliver a test build

    Verify signing settings, archive the build, and distribute it through TestFlight.

For CI/CD teams

Give automated builds a dedicated Mac

Register the physical node as a self-hosted runner and use a consistent toolchain for checkout, dependency installation, testing, archiving, and export. Use the GUI for initial authorization, signing checks, and troubleshooting; let the runner handle routine pipelines in the background.

Before running overnight batch builds, limit concurrency so multiple simulators and archive jobs don’t exhaust unified memory. Store signing keys in a controlled secrets store, never in the repository. Clean up temporary artifacts after each build, and leave the next user failure logs and change notes.

Don’t replace your production pipeline on the first run. Validate one app, one build target, and one signed export first, then migrate other jobs incrementally.

Runner setup checklist

  • Pin the Xcode version and Command Line Tools path
  • Check that Ruby, CocoaPods, and project lockfiles are consistent
  • Limit signing credentials to the build jobs that need them
  • Keep first-run logs, duration, and artifact verification results
For model experimentation teams

Run an observable inference session on Apple Silicon

Store quantized model weights on the remote drive and use MLX or llama.cpp to run inference experiments with small to mid-sized models. Run scripts over SSH, then review terminal output, Activity Monitor, and result files on the remote desktop. You can also share the GUI session when you need to present results.

Estimate memory pressure from weight size, context length, and concurrency—not just the model name. XV Studio M4 comes with an M4, 24GB of unified memory, and a 512GB SSD. For experiments involving multiple sets of weights or larger datasets, consider the optional +1TB SSD when ordering. More storage does not mean more unified memory.

Start with a small sample to check loading, time to first token, and sustained inference resource use before extending your rental. Whether a model fits in memory depends on the quantization format and runtime settings.

Record these four things first

Weights
Quantization format and file size
Context
Target length and batch size
Memory use
Free unified memory after loading
Results
Output path and reproducible experiment settings
Cross-time-zone handoffs

Take turns using the same development Mac

Let team members in different time zones connect to the same physical Mac during agreed time slots. Before signing off, the previous user should commit their code, log out of the desktop, and note the current branch and any unfinished builds. The next user can read the handoff notes before starting a session. Share scheduled access to the machine—not one person’s login.

A handoff note should include

  • Current branch, latest passing build, and artifact location
  • Running jobs, unresolved errors, and next steps
  • Next user’s access time and who will stop long-running jobs

Give each team member a separate system account with role-based permissions. Share credentials through a controlled channel, not a shared notes file. Avoid having two people control the same GUI desktop at once.

Migration steps

Move the project, align the tools, then connect the pipeline

Start with a small migration you can roll back. Don’t change the operating system, Xcode, dependency management, and build entry point all on day one.

01 / Data

Move projects and certificates separately

Commit and verify the code repository first, then use rsync or an encrypted archive to transfer project files that aren’t in version control. Transfer signing certificates and profiles separately with restricted access; never bundle private keys in a regular project archive.

Check:The repository checks out, all required files are present, and sensitive materials are absent from commit history.
02 / Toolchain

Pin reproducible versions

Record the Xcode version used by the project and check Command Line Tools, Homebrew dependencies, Ruby, and CocoaPods. Install dependencies and run a local build on the remote Mac first; address version differences one at a time.

Check:The remote Mac builds the same commit and passes tests on the target simulator.
03 / CI setup

Test-run a single pipeline

Register a self-hosted runner and configure repository access and signing keys with least privilege. Run a test build first, then archive and export. Record job duration, failure logs, and artifact locations.

Check:Repeat the initial run without needing ad hoc manual steps.
Choose by workload

Start with what runs concurrently, then choose memory and storage

Both configurations are dedicated physical Macs and support screen sharing, VNC, and SSH. Use your actual projects and model workloads as the basis for your choice.

Recommended M4 configurations by workload
WorkloadRecommended configurationHow to choose and what to check
Light builds and remote demosXV Desk M4
M4 · 16GB · 256GB
Good for editing a single project, routine builds, and demos. First check that your project, dependencies, and build artifacts fit in the available disk space.
Everyday development and simulatorsXV Desk M4 or XV Studio M4Start with Desk for a single target. For multiple simulators running in parallel, frequent archiving, or larger projects, consider Studio’s 24GB of unified memory.
Quantized model inferenceXV Studio M4
M4 · 24GB · 512GB
Optional +1TB SSD
Check memory requirements for weights and context first. Add extra storage only if you need space for multiple sets of weights.
Creative work and demos

Edit rough cuts remotely, but know the limits of streaming

For a quick demo asset, use Final Cut Pro on the remote Mac to organize clips, rough-cut a timeline, and export a preview. To showcase a macOS app, share the remote desktop so clients can see the workflow and interface in action.

Remote desktop frame rate, image quality, and responsiveness depend on client settings and network conditions. It works well for rough cuts, workflow demos, and reviewing results. For frame-accurate decisions, live color grading, or strict audio/video sync checks, use a local display and playback environment.

Next step

Take your workload to the plans page and choose before you order

For light builds, start with XV Desk M4. For multiple simulators or model inference, check XV Studio M4 first. Choose a node, rental period, and storage add-on to match your needs.