01 Define the Use Case
Distinguish Xcode CI/CD, iOS or macOS builds, multi-version testing, Unity iOS exports, and MLX inference. Your use case determines the priority of unified memory, storage, and runtime.
Start by defining your workload, then check the model, node, and term. After ordering, retrieve your connection details, check the environment, and run a verifiable Xcode build or MLX model-loading task.
No complex documents are needed. Define your use case, budget, target node, and local connection requirements to avoid switching repeatedly between models and terms.
Distinguish Xcode CI/CD, iOS or macOS builds, multi-version testing, Unity iOS exports, and MLX inference. Your use case determines the priority of unified memory, storage, and runtime.
Your budget should cover the base model, rental term, and storage add-ons. Do not compare daily prices alone; check the catalog amount for your planned term.
Schedule temporary validation by day or week; for ongoing builds and inference services, compare monthly or quarterly billing. Each term has its own catalog price—do not convert between them.
Choose Singapore, Japan (Tokyo), South Korea (Seoul), or Hong Kong. Prioritize your team’s time zone, data-transfer path, and the live availability shown in the console.
Account for source code, dependency caches, build artifacts, model files, datasets, and logs. If base storage is insufficient, add +1TB SSD or +2TB SSD when ordering.
Confirm a stable local network and prepare an environment for accessing the macOS graphical interface or command line. Retrieve connection details only through order management, not from public pages.
This timeline describes the pace of the actions needed to complete an order, not a fixed delivery commitment. Configuration, nodes, payment gateways, and connection details depend on order status and the live console response.
Choose one of VMCache M4 16, VMCache M4 24, or VMCache M4 Pro 64 based on memory, base storage, and task type.
Choose Singapore, Japan (Tokyo), South Korea (Seoul), or Hong Kong, then check the applicable catalog price by day, week, month, or quarter.
Choose USDT-TRC20, or use Visa, Mastercard, or Amex processed by Stripe. All orders are settled in US dollars (USD).
Return to the console after payment to check your order status. When connection details are available, follow the on-page instructions and save them in your team-approved secrets manager.
All three are dedicated physical nodes, not virtual machines. Each rental provides one dedicated physical machine, with no shared processor, unified memory, or base storage.
M4 · 16GB · 256GB
M4 · 24GB · 512GB
M4 Pro · 64GB · 2TB
Singapore, Japan (Tokyo), South Korea (Seoul), and Hong Kong operate continuously 365 days a year. Actual availability at ordering time is determined by the live console response.
Daily, weekly, monthly, and quarterly terms have separate catalog prices. Determine the actual usage duration first, then read the corresponding column. Do not multiply the daily price or round decimal amounts.
| Model | Configuration | Daily | Weekly | Monthly | Quarterly |
|---|---|---|---|---|---|
| VMCache M4 16 | M4 / 16GB / 256GB | $19.8 | $53.4 | $98.8 | $268.7 |
| VMCache M4 24 | M4 / 24GB / 512GB | $39.2 | $105.9 | $196.1 | $533.4 |
| VMCache M4 Pro 64 | M4 Pro / 64GB / 2TB | $60.3 | $162.9 | $301.7 | $820.6 |
For one-time validation, choose the term closest to the actual number of usage days; for continuous pipelines, choose monthly or quarterly billing based on the full operating plan.
Extra storage and Thunderbolt 5 aggregation have separate term prices. Add them only when the task requires them.
Before submitting payment, recheck the model, node, term, and add-ons, and confirm that the displayed USD total matches your expectations.
VMCache supports USDT-TRC20 and Visa, Mastercard, and Amex processed by Stripe. Actual gateway availability is determined by the backend response.
Pay using the network and amount shown on the order page. Confirm the network type and recipient details before sending; do not switch to another network.
Card payments are processed by Stripe. Whether this gateway is shown and available depends on the live order-page response.
After first access, establish an environment baseline before syncing source code, dependencies, or models. This separates connection, system, and task issues during troubleshooting.
Check the order ID, model, and node in the console to ensure you are connecting to the dedicated physical machine assigned to this rental.
Retrieve connection details only through order management. Never place passwords, private keys, or complete connection information in public tickets, repositories, or chat logs.
Confirm that the macOS version matches the task requirements. For multi-version testing, record the target version and test batch first.
Test reachability to code repositories, dependency sources, and internal team services. If connection problems occur, first distinguish local-network symptoms from node-side symptoms.
Check available space before downloading dependencies, datasets, or models, and leave room for caches, temporary files, archives, and logs.
Use clear directories for source code, caches, artifacts, models, and logs. Do not treat the node as your only backup location.
The first task is not about using all available resources immediately. Confirm that the toolchain, inputs, outputs, and log paths are reproducible. The two workflows below cover development builds and MLX inference.
Record the Xcode and command-line tools versions and ensure they match the project requirements.
Pull the source code and restore dependencies while monitoring cache directories and disk-space changes.
Run one build from a clean state and record the command, duration, and failure stage.
Confirm that the archive or test results are in the expected directory and retain redacted logs.
Confirm that the model files, tokenizer, and runtime script are compatible with the current MLX environment.
Load one model and record unified-memory usage; do not use a single peak as a substitute for long-term assessment.
Start the service on a controlled address and confirm access controls and log-redaction policies.
Start with one request, gradually increase batch size and concurrency, and record errors and loading logs.
The input source is clear, the command is reproducible, the output location is defined, logs are redacted, and the team knows which order and environment details to provide when issues occur.
Confirm the three models, four nodes, and four billing terms in the order flow. If you need to compare memory, storage, and total pricing first, return to the plans page and review each item.