What work would you like to finish during this period?
Describe a concrete result: a validated batch of images, an exported scene, a set of reviewed answers, or a finished computation. Specify who will check that result and by what date. If you are renting to explore a tool, set a learning goal, such as completing a full run and saving its settings. That is enough to give a first session some direction.
Gather a representative sample before ordering. The files should open on your machine and come with the necessary supporting items: linked assets, the configuration or model used. For a small team, designate one person responsible for this set. That way you avoid starting the period waiting for data or a decision scattered across several colleagues.
Is your software compatible with the GPU you have in mind?
Note the exact version of your tool and check its prerequisites. Verify its GPU engine, the required libraries, and your project's extensions. A program built for CUDA should not be assumed to be interchangeable with a ROCm installation. Vendors' compatibility matrices let you check which combinations are supported before choosing an environment setup.
Keep a short startup procedure: install or open the software, load a sample, select the GPU, then produce a small output. Choose the setup offered in the configurator based on that workflow. Ubuntu, PyTorch, or Blender serve different needs; your choice should follow the application you will actually use, not just the card name that feels familiar.
How much memory and how many units do you need?
Note your project's known requirement and how it was established: the software's prerequisites or a measurement from a test run. Then compare the 24, 48, or 80 GB models where they match that requirement. Memory per card is the benchmark to follow for a task run on a single GPU. The size of the folder on disk does not let you directly infer that usage.
Also check the number of cards in the unit. The B200 is offered in units of two GPUs; the other units in the catalog contain one card. Multiple GPUs can be used to spread out independent tasks or to run an application designed for them. Their memory does not automatically add up for ordinary software. Start with a quantity your work plan can actually use.
When will you work, and where will you retrieve the results?
Choose 3, 7, or 30 days based on the time slots you have available and your validation pace. For a short assignment, allow time for setup, testing, the final work, and retrieval. If a client needs to respond between two steps, factor that delay into your planning. Your schedule is often a better benchmark than the estimated duration of a single run.
Prepare the destination for your results before you start. Check that you can copy a file there and that you have enough space. Decide what you will keep along with the deliverables: settings, versions, file mapping, and any useful resume states. The last time slot in your period should leave you enough time to open the copy you retrieved.
How does ordering work without KYC?
In the BriefGPU configurator, select the GPU, duration, units, and setup. For a first order, create your account with first name, last name, email, and password. If you already have an account, log in; your details are then pre-filled. The process does not require an ID document or KYC verification. Review these details and the total in USD before confirming; they go with the record you can view in your account.
Your account lets you find your orders again from another browser by logging in with your email and password. Keep your order reference to make it easier to track. An email address alone is not enough to open orders. Also prepare the wallet or tool you will use for payment so you can check which networks it supports.
What should you do after sending the crypto payment?
Choose the method shown in your order, then use the network, address and exact amount displayed before the deadline. The copy buttons save you from copying these values by hand. After sending, click "I've paid". The file keeps your report and shows its tracking; this click sends no funds and does not itself confirm the settlement.
If you have several payment requests in your history, check the one that matches your transfer. The payment and the preparation of the rental stay readable in their respective states. To request help, provide the order reference and the action concerned. A clear description of the problem is enough; no private key or recovery phrase is needed.
Example: starting with a ready session file
A freelancer wants to test an assistant on twenty questions over 3 days. Before ordering, they gather their permitted documents, choose a compatible tool, prepare the expected answers and book a slot for the evaluation. They keep an output file for the results and the settings. Once the rental is provisioned, the first task is already defined. They keep the choice of their software and their processing; BriefGPU does not inspect the contents of files, prompts or computations.