Mini S
M4$99.6/mo · works out to $3.3/day
- 16GB unified memory
- 256GB NVMe SSD
- 1Gbps dedicated bandwidth
- Available in all 5 regions
From the M4 16GB to the M4 Pro 64GB — specs, pricing, and stock, all on one page. By the time you finish reading, you'll know exactly what renting one costs, and what not renting one costs you.
Every tier is a fully dedicated hardware machine — one physical Mac mini serving only you. The only differences are chip, memory, and SSD.
$99.6/mo · works out to $3.3/day
$198.4/mo · works out to $6.6/day
$302.7/mo · works out to $10.1/day
| Plan | Singapore | Japan (Tokyo) | South Korea (Seoul) | Hong Kong | US West (Silicon Valley) |
|---|---|---|---|---|---|
| Mini S | |||||
| Mini MM4 · 24GB | |||||
| Mini ProM4 Pro · 64GB |
From the $19.9 Mini S to the $60.5 Mini Pro, the delivery standard is identical. Every region runs 365 days a year, with no scheduled downtime.
Full admin access delivered from day one — install any version of Xcode, Homebrew, or kernel extensions without waiting on approval.
Out-of-band graphical access — get into the desktop to troubleshoot even if the system hangs or SSH drops. The full macOS GUI is always available.
Trigger a full reinstall from the console with one click — back to factory state in about 25 minutes, no ticket, no extra charge.
Before you cancel, package your home directory into a compressed image and take it with you — move to a new machine or your local disk without losing your setup.
1Gbps dedicated bandwidth with no data caps — uploading a 1GB .ipa or pulling 50GB of model weights costs exactly the same.
If uptime falls short, you're credited 3x the affected downtime. See the Trust page for the credit terms and daily status history.
Q1
Go with Mini S (M4 16GB). A cold xcodebuild of a mid-sized Swift project takes about 4–6 minutes for a single project, and a full fastlane build-and-upload cycle finishes within 15 minutes. 16GB is plenty for a single pipeline.
Q2
Start at 24GB (Mini M / Mini Pro). With 2–3 concurrent jobs, each Xcode compile process can peak at 6–16GB; past 3 concurrent jobs, or if you need a simulator matrix, go straight for the 12-core M4 Pro.
Q3
Go with Mini Pro (M4 Pro 64GB). MLX loading a 70B 4-bit quantized model uses roughly 40GB of unified memory; models under 32B run fine on Mini M, but leave very little headroom for the system.
Not sure? Rent a Mini M for a day and test it yourself — run your own CI scripts, watch memory pressure in Activity Monitor, then decide whether to size up or down. Switching tiers never wastes time you've already paid for.
Upgrades, downgrades, and refunds are all self-service in the console. For edge cases (like a cross-cycle upgrade), email support@oncemini.com or open a ticket — we respond within 2 hours during business hours.
$ rent --plan mini-m --days 1
Test-drive a Mini M for $39.7 a day — run your build scripts before locking in a longer cycle. SSH/VNC credentials arrive in about 5 minutes on average, and your first order is fully refundable within 24 hours.
Pick your region at checkout — actual availability at order time is returned live by the portal. See Regions for latency benchmarks and Support for the getting-started guide.