Six categories of frequent questions — billing, delivery, permissions, CI, networking, and compliance — each answered with concrete numbers and edge cases. Read this page and you probably won't need to open a ticket.
Start times, refund calculations, and renewal grace periods — all spelled out down to the hour.
Billing starts from the timestamp on your delivery email, counting one day per 24 hours — not by calendar day. Example: if you receive your credentials at 2:00 PM Wednesday and rent for 3 days, your rental ends at 2:00 PM Saturday. Orders queued for delivery don't accrue rental time before delivery; the wait between payment and delivery isn't counted.
All three tiers offer daily, weekly, monthly, and quarterly terms — the longer the term, the lower the per-day rate. Comparing the two extremes: Mini Basic runs $21.5/day or $107.7/month; Mini Pro runs $59.3/day or $296.7/month. Pick daily for one-off verification runs, or go monthly for a long-running CI runner to get the better rate. Full pricing tiers are on the rental plans page.
You can request a refund within 168 hours (7 days) of delivery, prorated by unused days: the used portion is deducted at that tier's daily rate (any partial day counts as a full day), and the remaining balance is refunded to your original payment method. Orders past the 7-day window are no longer eligible for a refund, but you can keep using the rest of your rental period as normal. To request one: Dashboard → Orders → Request Refund, or email support@armmini.com.
No. You'll get a reminder email 72 hours before expiration, plus a 48-hour renewal grace period after expiration during which your machine and data stay exactly as they are — renew and you continue seamlessly. If you still haven't renewed once the grace period ends, the machine is taken offline and released, and the disk is wiped immediately and irreversibly. If you're running a long-lived runner, we recommend monthly or quarterly billing to cut down on renewals.
We accept USDT-TRC20, as well as Visa / Mastercard / Amex (processed via Stripe) — all charged in US dollars (USD). Available gateways may vary; check the checkout page for what's currently offered. After each successful payment, you can download the electronic receipt from the billing page in your dashboard, including order ID, tier, term, and amount.
From payment to your first desktop connection, the standard path is just one email away.
Tiers are delivered within 10 minutes of payment in most cases, and within 24 hours at peak — the wait isn't counted toward your rental period. Note: if payment isn't completed within 30 minutes of placing an order, the order is automatically canceled and you'll need to reorder. Per-node delivery details are shown in the node matrix on the plans page.
Via encrypted email to the address used at checkout, including your dedicated IPv4 address, VNC port and initial password, and SSH account. We recommend doing two things right away: change the initial VNC password, and switch SSH login to your own public key with password login disabled. Exact commands are in the connection guide's security hardening checklist.
By default only two inbound ports are open — VNC on 5900 and SSH on 22 — with everything else blocked. Both can be adjusted yourself in the dashboard firewall: change the port number, restrict source IPs, or disable VNC entirely and keep only SSH. Before exposing any service publicly, we recommend tightening the allowed source IPs in the firewall first.
A clean macOS admin account, with commonly used versions of Xcode and the command line tools preinstalled, Screen Sharing and Remote Login already enabled, and no monitoring agents or extra software installed. The entire physical Mac mini belongs to your account alone — any previous tenant's data was fully wiped when their rental ended.
The whole machine is yours, and so are the permissions.
Yes. The delivered account has full sudo access — install Homebrew, any dev tools, change system settings, manage login items, no different from the Mac on your desk. Both the macOS GUI and command line are fully usable. The only boundary is the acceptable use policy in our Terms of Service: no spam, network attacks, or other prohibited uses.
Yes. At checkout you can pick the release version of macOS 14 (Sonoma) or macOS 15 (Sequoia), and you can switch versions later by reinstalling; beta builds aren't offered. Xcode ships with a commonly used version preinstalled — download additional versions yourself and switch between them with xcode-select.
Each billing period includes 1 free reinstall, which you can trigger yourself in the dashboard; it takes about 30 minutes, during which the machine is unavailable. A reinstall wipes the entire disk — everything on the SSD is unrecoverable, so use rsync to pull anything you want to keep back to your local machine first. Additional reinstalls beyond the free one can be requested via ticket and are handled on a scheduled basis.
Running the machine as a self-hosted runner? Get these right first.
Don't run the runner as a foreground process. Use GitHub's official svc.sh to install it as a launchd service, so it's automatically relaunched if the process exits or the machine reboots:
$ cd ~/actions-runner
$ sudo ./svc.sh install && sudo ./svc.sh start
$ launchctl list | grep actions.runner
ok actions.runner.armmini.mini-01Nodes run 365 days a year with no scheduled downtime; occasional disconnects are usually just public network jitter, and once it's running as a service the runner will automatically re-register as online. Same idea for GitLab Runner — install it as a service with gitlab-runner install.
It comes down to memory: a full Xcode build peaks at roughly 6~10GB, and each simulator adds another 1~2GB. As a rule of thumb: Mini Basic (16GB) handles 1 concurrent build; Mini Standard (24GB) handles 1~2; Mini Pro (M4 Pro, 64GB) handles 2~3 and can run multiple simulators at once for UI testing. Overloading concurrency just slows down every pipeline — adding another machine is usually the better move.
This is exactly where a dedicated physical machine beats hosted CI: the disk sticks around, so caches persist naturally. Do three things — pin DerivedData to a fixed path (e.g., ~/ci-cache/DerivedData); pass -derivedDataPath pointing to it in xcodebuild or fastlane; and remove the cleanup step from your pipeline. Incremental builds typically save 40%~70% of the time versus a full rebuild. SPM and CocoaPods caches work the same way — just pin them to a fixed path.
Bandwidth, addresses, and firewalls — the three most-asked network facts.
All five physical locations — Singapore, Tokyo, Seoul, Hong Kong, and Silicon Valley — have a 1Gbps shared uplink with unmetered traffic under a fair-use policy. Day-to-day work like pulling dependencies, pushing images, or uploading an IPA to App Store Connect doesn't require watching a usage meter; sustained abuse that saturates the uplink will be throttled with a notification.
Each machine gets 1 dedicated IPv4 address, not shared with anyone, and it stays fixed for the entire rental period. That means you can safely add it to your company firewall allowlist, a CI platform's static egress list, or a private dependency registry's access policy — the IP won't change when you renew.
Dashboard → My Instances → Network → Inbound Rules — you can add, edit, or delete allow rules yourself, and changes take effect within about 1 minute. Rules can be scoped by port, protocol, and source IP range. Our suggestion: allow SSH only from your office or home broadband's egress IP, and keep VNC off the public internet entirely, tunneling it through SSH instead.
It depends on your location and ISP: the Tokyo node typically sees 30~60ms, Seoul 30~50ms; users on the US West Coast see about 25~35ms to the Silicon Valley node, and the Singapore and Hong Kong nodes offer similarly low regional latency for Southeast Asia and southern China. A full ping table broken down by city and carrier (including test dates and sample sizes) is in the connection guide's latency benchmarks — worth a glance before you pick a node.
Eight terms that keep coming up on this page, each explained in a sentence so we're all on the same page.
Who owns the data, and how wiping is proven — spelled out in writing.
Your data belongs to you, full stop. Our operations team never logs into your machine's OS and never views or copies your data; routine monitoring only collects out-of-band hardware metrics — power state, network port status, and data center environmentals — nothing on the SSD is ever touched. Full disclosure is in our Privacy Notice.
Within 7 days of account closure, every associated machine gets a full disk wipe, and we generate a wipe record — including machine serial number, wipe timestamp, and method — available on request via ticket. Account and billing logs are retained for 90 days per our privacy notice before deletion. Retired disks are always physically destroyed, never resold.
No — backups are on you, which is really the flip side of our "we don't peek at your data" promise: since we can't access your files, we can't back them up either. We recommend using rsync to regularly pull build artifacts and configs back to your local machine, or uploading key CI artifacts to your own object storage. Always double-check your backups before a reinstall, closing your account, or letting a rental lapse.
Two paths, depending on your issue: troubleshoot it yourself with the guide, or open a ticket if you need a human.
VNC black screen, SSH timeout, Xcode signing failures — our troubleshooting guide gives you commands step by step by symptom, and most issues resolve in under 10 minutes on your own.
Billing disputes, bulk purchasing, migration assistance, and other one-off cases: log into the dashboard and open a ticket (first response <2h), or email support@armmini.com.
VNC/SSH credentials are emailed within 10 minutes of payment, backed by a 7-day refund window — try it for a day before you commit.