Runner Types
By default each test run uses the following underlying hardware specifications:
| Runner Type | Platform | Host OS | Architecture | RAM | CPU Cores |
|---|---|---|---|---|---|
default | iOS | MacOS | M1 | 8 GB | 8 |
cpu1 | Android | Ubuntu | x86 | 15 GB | 13 |
Android runs use cpu1 unless you choose another runner type. Default iOS runs usually run on M1 hosts, but an idle M4 host can also pick them up.
Simulators and emulators keep needing more resources, so we offer extra runner types for heavier apps and tests.
INFO
The m4 runner type is in beta while we add capacity.
If your app or test struggles on the default runner types, try the options below. m4 and m1 runs are charged at the advanced rate, see our pricing.
Available runners
| Runner Type | Platform | Host OS | Architecture | RAM | CPU Cores | Notes |
|---|---|---|---|---|---|---|
m4 | iOS | MacOS | M4 | 16 GB | 10 | |
m1 | Android | MacOS | M1 | 8 GB | 8 | Apple Silicon (arm64). Google Play runs are moving here, see below. |
WARNING
The gpu1 runner type was removed on 9 November 2026. Android runs that still pass --runner-type gpu1, or set it with DEVICECLOUD_OVERRIDE_RUNNER_TYPE, run on cpu1 instead, billed at the standard Android rate. They don't fail. On iOS, gpu1 is still rejected. Stop passing gpu1: Android runs use cpu1 by default, or you can pass --runner-type cpu1. See the Support Policy for more information.
INFO
Google Play runs are moving from cpu1 to m1, at the same price. See Deprecations & Removals for the date.
Limitations
m4is iOS only. Choosing it for an Android app is rejected.cpu1andm1are Android only. On iOS,m1is treated asdefault(with a warning), andcpu1is rejected.- On
m1, every flow must run on API level 34 to 37. This includes devices chosen per flow or in a device matrix. - Google Play is available on
m1onpixel-8(API 34) andpixel-10(API 36, 37). See Google Play APIs. m1runs arm64-v8a code only. An APK with native libraries must include arm64-v8a ones, otherwise the run is rejected.- Generic Tablet isn't available on
m1, so don't select it with that runner type.
Usage
To use a specialised runner, use the CLI flag --runner-type.
dcd cloud ... --runner-type=<type>
# For example, to use the m4 runner:
dcd cloud ... --runner-type=m4Overriding the runner type for a single flow
To run just one flow on a different runner type, set DEVICECLOUD_OVERRIDE_RUNNER_TYPE in that flow's env: block. It overrides --runner-type for that flow only. The other flows in the upload keep the upload's runner type.
# in your test.yaml
appId: my.app
env:
DEVICECLOUD_OVERRIDE_RUNNER_TYPE: m4
---
# test steps- The same limitations apply as for
--runner-type. The one difference: on iOS,m1becomesdefaultwithout a warning. On Android,defaultbecomescpu1. - A value that isn't a runner type, or one that the platform doesn't support, rejects the whole upload before anything runs. The flow is never quietly run on the upload's runner type instead.
- The flow is charged at its own runner type's rate. For example, an
m4flow in adefaultupload is charged the advanced rate, and the other flows are charged the standard rate. - In a device matrix, the flow still runs on every device in the matrix, each time on its own runner type.
- Like all per-flow overrides, this only works in the flow's
env:block, not with-e.
Runner Type Support Policy
How we retire runner types is set out in our Support Policy.
