Icemoon
← Home
Guide
iPhone farm software
Last updated: August 1, 2026
The hardware side of a farm is a solved problem — phones, a hub with enough amperage, a shelf. The part that decides whether the thing works is the control layer: the software that actually drives the screens. There are five realistic options, and most of the noise online conflates them. Here is the honest map.
The five control layers
| Approach | Runs on current iOS? | Fleet features | Real cost |
|---|---|---|---|
| Build on WebDriverAgent yourself | Yes | None — you write them | Engineering months, then ongoing maintenance every iOS release |
| Appium | Yes | None — it is a control channel, not a farm | Free, plus the scheduler and dashboard you build on top |
| Jailbreak tooling (XXTouch Elite, AutoTouch) | No — needs a jailbreak; the TrollStore path stops at iOS 17.0 | Per-device install and management | $9.99 once per device, plus hardware frozen on old firmware |
| Apple Shortcuts | Yes | None | Free — but only automates what apps choose to expose |
| Android cloud phones (GeeLark, Multilogin, VMOS) | Not applicable — no iOS at all | Extensive, on Android | Metered device time and proxy traffic |
| Icemoon | Yes, iOS 16+ | Dashboard, scheduler, per-device accounts and proxy, script library | $39 / $99 / $249 a month, flat per fleet |
Two of those rows are the ones people actually get wrong. The cloud phone platforms are marketed as the modern answer to phone farms, and they are — for Android. Not one of them offers an iOS environment. And jailbreak tooling is described as working "without a jailbreak" on the strength of a TrollStore build, which is true only up to iOS 17.0, because Apple closed the underlying CoreTrust flaw in 17.0.1. The details are here.
What a fleet needs that a single device doesn't
Automating one iPhone and automating thirty are different problems, and the second one is mostly not about touch input. The things that break at scale:
Identity per device. Each phone needs its own accounts and its own network egress, assigned and remembered. Doing this by hand across thirty devices is where most farms quietly fail — iOS attaches proxies to Wi-Fi networks rather than to devices, which is less convenient than it sounds.
Scheduling, not scripting. A flow you have to launch manually is not automation at fleet scale. Work has to start on a calendar, across devices, without you present.
Failure that is visible. When one phone in a rack stops mid-flow, you need to know which one and why. Without a dashboard you find out days later from the output.
Input that is not identical every time. Replayed coordinate scripts produce the same gesture with the same timing on every run and every device. Generating each swipe as its own curve is a different mechanism — we published the maths rather than asserting it, along with what session timing looks like when it is machine-generated.
Power that holds. Not software, but it presents as software. A hub that cannot supply enough current under load produces disconnects that look exactly like automation bugs — the amperage arithmetic is worth doing before blaming the control layer.
Where Icemoon fits
Icemoon is the fleet layer, built for the case where the devices are real iPhones you own and they need to stay on current iOS. It runs on a Mac (Apple Silicon); phones connect over USB and keep taking OS updates, because nothing about the device is modified. Flows are built by recording a task once, by assembling them visually, or by describing the goal to the built-in AI copilot, which reads the screen and executes through an MCP server.
What it is not: it is not a cloud service, so there is no hardware to rent and nothing to spin up elastically. It is not a bot catalogue — there are no ready-made flows, you build your own. And it does not run untethered: the Mac is required.
Pricing
Flat per fleet, no metering: Solo $39 for one device, Farm $99 for five, Scale $249 for twenty. Yearly billing works out to $31 / $79 / $199 a month. Devices running around the clock cost the same as devices running an hour a day. Payment is in USDT (TRC-20) or SOL, and nothing auto-renews.
FAQ
What software do you use to run an iPhone farm?
Five realistic control layers: a self-built stack on WebDriverAgent, jailbreak tooling such as XXTouch Elite, Apple Shortcuts, an Android cloud phone service (which cannot drive iPhones at all), or a purpose-built fleet controller like Icemoon. Which fits depends on how many devices you run, whether they must stay on current iOS, and how much engineering time you are willing to spend on the control layer itself rather than on the work it does.
Can you automate iPhones without a jailbreak?
Yes. A signed runner built through Apple's own toolchain drives a stock iPhone over USB from a Mac. The device keeps taking iOS updates and nothing in its security model is disabled. This is how Icemoon works, and it is the reason current hardware is usable at all — public jailbreaks for recent iPhones on current iOS are effectively nonexistent.
How many iPhones can one Mac drive?
In practice the limit is power and USB topology, not the Mac. Plans are sized at 1, 5 and 20 devices. Past roughly ten phones the binding constraints become hub amperage and cable quality, which produce faults that look like software failures and are not — the arithmetic is here.
Is Appium enough to run a farm?
Appium gives you a control channel, not a farm. No scheduler, no per-device account or proxy assignment, no fleet dashboard, and input that is exact rather than human-shaped. Teams who run farms on Appium end up building that layer themselves, which is a real project rather than a weekend of configuration — the longer comparison goes into where each one genuinely wins.
Which iPhone models work for a farm?
Anything on iOS 16 or later — in practice iPhone 8 and newer, which is 32 of the 36 models released to date. Building around a single screen class pays off: 390×844 points is shared by six different models, so coordinate-based flows move between them without edits. The full model table lists every ceiling.