Rent an H200 NVL
NVIDIA H200 NVL
Specifications
Renting a H200 NVL
Rent an H200 by the minute
The H200 NVL carries 141 GB of HBM3e on one card, more than five RTX 4090s put together, with memory that moves data about four times as fast as theirs. It is the card for a model that does not fit anywhere smaller, or for a mid-size model served to many users with very long prompts.
On ParalonCloud a machine is rented whole and billed by the minute. A rental opens as Jupyter Lab in your browser, with a terminal; the hourly rate you see when you start is locked for the session, billing stops the minute you stop, and there is no contract or minimum spend.
This card is rare on the network. The top of this page shows whether one is free right now, and the list price when none is. The figures below were measured by us on an H200 NVL in a rental, the same way a customer would reach it.
What fits
What fits in 141 GB
On a card this size the question is what is left for context after the model has loaded, and how many long requests it can hold at once.
Measured
Measured on an H200 NVL
From a rental on 11 October 2026, from the card's own verification, and from serving a 27B model on it the same day. The spec sheet gives 4,800 GB/s of memory bandwidth.
Steps
How to rent one
- 01
Check availability
The top of this page says whether a machine with this card is free. If none is, the rent page lists what is, including the 96 GB RTX PRO 6000 and pairs of smaller cards.
- 02
Start the rental
Choose Jupyter Lab, accept the terms and start. You need credits for one hour to open a session and are charged only for the minutes you use.
- 03
Work, then stop
Use the terminal for pip, conda or apt, run your job, download what you want to keep, and stop.
Limits
Know before you rent
- This card is rare on the network and is often not available. The price shown when none is free is the list price, not an offer.
- The largest open models do not fit on one H200. One card is for a model whose weights and context together stay under about 128 GB.
- A rental is an unprivileged container: Docker inside it does not work. Install what you need with pip, conda or apt.
- Access is through Jupyter Lab and its terminal. Other ports are not exposed to the internet.
- Everything on the machine is wiped when the rental ends. Download results before you stop.
FAQ
Questions about the H200 NVL
How much of the 141 GB can I use?
The verification allocated 128,367 MB of the 143,157 MB the card reports. The rest is held by the driver and by the system the card sits in.
Is the card running at full speed, or is it capped?
In our rental the power limit was 600 W, the card's default, and the PCIe link was at its maximum. Memory bandwidth measured 4,062 GB/s in verification and up to 4,113 GB/s with PyTorch, about 85% of the 4,800 GB/s on the spec sheet.
Why did this card once show 48% of spec?
Our old bandwidth test used one GPU thread per number, which is limited by something other than memory on a card this fast. It read 2,319 GB/s. We rented the card, measured it another way, replaced the test, and the card now reads 4,062 GB/s. The full account is linked from this page.
When is an H200 worth it over smaller cards?
When the model or the context does not fit in 24, 32 or 96 GB, or when you serve many long requests at once. For one user and short prompts a single RTX 4090 or RTX 5090 costs much less.
Is it available to rent right now?
The badge at the top of this page says so, from the live network. Machines with this card come and go; when none is free the page shows the list price and the rent page shows what is available instead.
Read next
Need H200 servers for months rather than hours? Reserved H200 capacity from data centers.