First confirm your order, node, and access credentials, then determine whether the issue involves your local network, remote session, macOS, or workload. Follow this step-by-step checklist to record findings and send them to the support team.
Change only one setting at a time. After each step, record whether anything changed. This narrows the cause and keeps the reproduction path in your ticket clear.
The delivery details should match the model, term, and node selected in your order. If anything differs, save screenshots and the order number first. Do not reinstall the system yourself.
Confirm that the order shows the selected node among Singapore, Japan (Tokyo), South Korea (Seoul), Hong Kong, and the US West. Network tests must target the actual node in the order; results from another region cannot be substituted.
MB M4 16 should show M4, 16GB RAM, and 256GB SSD; MB M4 24 should show M4, 24GB RAM, and 512GB SSD; MB M4 Pro 64 should show M4 Pro, 64GB RAM, and 2TB SSD.
Copy the access address, username, and initial credentials from the order details instead of transcribing them manually. After the first successful connection, update the initial credentials immediately and store the new credentials in a team-approved controlled location.
Record the current macOS and Xcode versions before installing project dependencies. CI/CD teams should add version numbers, the command-line tools path, and dependency lock files to the environment baseline so individual tasks do not change the toolchain.
Confirm the screen-lock policy, credential storage location, remote-session sign-out method, and team access scope. Never share access credentials through public documentation, group-chat screenshots, or build logs.
Record the macOS, Xcode, shell, package manager, and project dependency versions, and save initialization commands that can be run repeatedly.
Before investigating connection lag, check CPU, memory, disk space, and long-running tasks to distinguish node load from network experience.
Save a version inventory before upgrading the toolchain. If an issue appears, verify changes in reverse order, starting with the latest; do not reinstall multiple components at once.
Being able to open the access page in a browser does not mean the remote desktop is connected. Seeing the desktop but being unable to type does not necessarily mean the network is down. Check by symptom to avoid ineffective retries.
Return to the order details and recopy the access address and username, checking for spaces before or after the values. If a team member recently changed the credentials, stop retrying the old ones to avoid triggering security restrictions. If authentication still fails, include the order number, failure time, access method, and full error message in your ticket. Do not put the credentials themselves in the ticket.
Retest first over a wired connection or stable Wi-Fi. Pause VPNs, proxies, and large uploads, then record the interval between disconnections. Compare with another supported browser, but do not establish multiple sessions at once. If switching networks resolves the issue, record the local carrier, packet loss, and round-trip latency.
Wait for the current session to finish loading its first frame, then resize the browser window. Confirm that the browser is not blocking graphics rendering, and check whether the node is running a high-load build or inference task. If the display remains blank while the connection status is normal, record the browser version, black-screen duration, and node load before submitting a ticket.
Check that the keyboard layouts on your local system, browser session, and remote macOS match. Test letters, numbers, symbols, and common shortcuts in a plain-text editor rather than judging directly in an IDE. When reporting the issue, specify the local keyboard type, expected keypress, actual output, and remote input method.
First test bidirectional copying with one short line of unformatted text, then check the browser's clipboard permissions. Do not transfer large text, images, or files through the clipboard. Use a code repository, controlled download URL, or team-approved secure transfer tool for project files.
99.9%
The service operates continuously throughout the year. Availability is calculated according to the service scope, measurement method, and exclusions defined in the service terms; your own network, credential actions, and force majeure are excluded from node downtime.
If the service target was not met during the measurement period, submit records under the service terms and request eligible service credits.
View the service terms“It’s slow” alone cannot distinguish a local link, cross-region routing, browser rendering, or node workload issue. Take at least three consecutive samples while keeping the target node and network environment unchanged.
Record the carrier name, access city, and whether you use wired networking, Wi-Fi, or mobile data. If you use a VPN or proxy, state whether you retested after disabling it.
Specify the Singapore, Japan (Tokyo), South Korea (Seoul), Hong Kong, or US West node actually used by the order. Do not list only the country or region.
Record the local time, time zone, and duration in minutes when the issue occurred. For intermittent issues, provide at least two time windows.
Sample at stable intervals and submit the total sample count and packet-loss percentage. Do not provide only a screenshot of one failed request.
Provide the minimum, median, maximum, and any clearly jittery intervals. Latency figures should correspond to the same packet-loss test round.
State whether the node was compiling, running a simulator, performing batch tests, or running model inference, and record an overview of CPU, memory, and disk space.
Orders are generated from the selected model, node, add-ons, and term. All amounts are charged in US dollars (USD); the gateways actually available are determined in real time by the console.
Rent by the day, week, month, or quarter. The period is calculated from the order record after confirmation and entry into the delivery process. Before renewing, verify the current model, node, add-ons, and new service period.
The bill shows the base model, term, node, and selected add-ons. If it does not match your order, retain the billing and order numbers and verify them through a console ticket before payment.
Only USDT-TRC20 and Visa / Mastercard / Amex (via Stripe) are supported. All amounts are charged in US dollars (USD); the status returned by the console after payment is authoritative.
Sign in to the console, open the relevant order, verify the current expiry details and renewal options, and complete payment. If you need to change the model, node, or storage expansion, first confirm that the new configuration fits your workload.
Do not submit repeated payments. Record the billing number, order number, payment method, submission time, and current console status, then request verification through a ticket. When sharing payment records, first mask unnecessary identifying information.
Support is available only through console tickets and support@minibinary.com. For existing orders, dropped connections, and node issues, use a ticket first so the order status and follow-up records can be linked.
For questions about configuration choices, node locations, renewal rules, and operating procedures. State the expected workload, number of concurrent tasks, and preferred node.
For intermittent disconnections, input issues, build-environment changes, or billing-status questions. Provide the frequency, scope of impact, and troubleshooting steps already completed.
For connection failures, persistent black screens, or critical task interruptions. Mark the ticket subject as “Usage impacted” and provide the order number, node, time of occurrence, and reproducible steps.
Capture the relevant lines before and after the error, keeping timestamps, error codes, and call locations. Before submitting, remove credentials, tokens, keys, personal information, private project URLs, and business data.
Do not upload a complete home directory, complete environment-variable set, or unscreened build archive. If you are unsure whether content is sensitive, describe the field type first and wait for the support team to confirm what is needed.
For existing orders, submit a console ticket first. For configuration, node, or bulk-deployment questions before ordering, email the purpose, desired configuration, term, and number of concurrent tasks.