For users in South China & Southeast Asia, run builds, tests, and inference workloads on a dedicated Apple Silicon Cloud Mac. Keep working on your local computer while the cloud node handles the heavy lifting.
From $20.2 per day. Choose a configuration and billing term, then review your order. Availability is confirmed in real time by the console.
The chip, memory, and base SSD are fixed for each configuration. Try a short workload by the day, then consider weekly, monthly, or quarterly billing for ongoing builds. Select a term below to see its price, then confirm your configuration and rental period on the order page.
Showing monthly prices. All prices are charged in USD.
Ideal for single-stream iOS builds, command-line tasks, and lightweight CI runners. Run a full build first to gauge your cache and storage needs.
$100.8
per month · USD
A dedicated physical Mac, not a virtual machine. Project files, build caches, and artifacts share the base storage, so we recommend regularly clearing anything you can rebuild.
Rent Flow M4Designed for parallel builds, larger test workspaces, and MLX quantization experiments. The models you can run depend on their weights, context, and runtime overhead.
$306.1
per month · USD
A dedicated physical Mac, not a virtual machine. More unified memory can accommodate more parallel workloads, but it doesn’t mean every model will run as-is.
Rent Flow M4 ProNeed more space for projects or models? At checkout, you can select +1TB SSD, +2TB SSD, or a Thunderbolt 5 expansion option. Add-ons are priced according to your selected billing term and aren’t included in the base plan prices above.
For users in South China & Southeast Asia, the regional average latency estimate is 20~40ms. Use it as a guide when choosing a node—not as a live test of your connection or a guaranteed latency for every session.
VNC responsiveness also depends on bandwidth, packet loss, resolution, and how much the screen changes. Start with a lower resolution and image quality, then increase them once interaction feels smooth.
A guide for interactive workloadsSSH commands and CI builds aren’t affected by desktop image quality. Dependency downloads, code pulls, and artifact uploads each use their own network paths, so a single latency figure won’t tell you how long a build will take.
Test with a real projectOffice networks, home broadband, and mobile hotspots can take different routes. Try the service from your usual work connection, note the time and connection type, and then decide whether a longer rental makes sense.
Note your network and test timeNodes run year-round, 365 days a year. Network latency and service availability are different metrics: latency describes the connection experience, while availability is calculated according to the service commitment. Neither can stand in for the other.
These timings are estimates for a typical setup. You can create an account and choose a plan in advance; payment confirmation, physical node allocation, and system readiness are reflected in your order status in the console.
About 1 minute. Create an account with an email address that can receive verification codes. Your orders and support requests are managed under the same account.
About 30 seconds. Select a model, node, and daily, weekly, monthly, or quarterly term. Add storage or an expansion option if needed, then check the order total.
Card payments are usually confirmed instantly; on-chain USDT-TRC20 confirmation takes about 1 minute. We accept only USDT-TRC20 and Visa / Mastercard / Amex processed by Stripe. Available payment gateways are confirmed by the backend. Before sending funds, check that you’re using the TRC20 network and verify the payment details.
Allocation and system readiness take about 1.5 minutes. Once you receive your VNC / SSH credentials, connect and change your password before installing or checking the toolchain versions your workload requires.
A dedicated machine doesn’t automatically back up your data. Before you start, decide where to back up your code, certificates, models, and build artifacts. Export anything you need to keep before your rental ends.
This section covers node selection and rental terms. For connection credentials, payment status, or an existing order, sign in to the console and submit a support ticket.
Start by considering who needs to use the desktop regularly, then look at how code and artifacts will be transferred. For automated builds, test dependency downloads and artifact uploads first. If several people need frequent remote access, each person should test from their own work network rather than relying on one person’s experience.
No. It’s a regional average latency estimate for Hong Kong and users in South China & Southeast Asia, not a measurement of your specific provider, route, or time of day. Your experience also depends on packet loss, jitter, and desktop encoding overhead. Test with a real workload before choosing a longer billing term.
If you haven’t tested your toolchain, network, and storage needs yet, start with a day and run the complete workflow from pulling code to exporting artifacts. Once you have a stable workload, compare weekly, monthly, and quarterly pricing against your expected usage. Switching terms here only changes the prices shown; confirm the final model, node, and term on the order page.
Choose a configuration, then confirm your node and rental term. Start with a full build or inference experiment, and move to a longer rental once you’ve validated your setup.