NVIDIA Hopper
Rent H100 servers, reserved from data centers
H100 is the Hopper GPU most AI clusters were built on, with 80 GB of HBM3 and FP8 support. It is the most widely available data-center GPU, which keeps its price falling as newer generations ship.
- Memory per GPU
- 80 GB HBM3
- Architecture
- Hopper
- Listings
- On request
- From
- Ask
H100 capacity listed now
No H100 capacity is listed right now. Send a request and we will source it from our owners.
Tell us what you needWhat H100 is good for
- Training and fine-tuning up to tens of billions of parameters
- Cost-efficient serving of 7B to 70B models
What to ask before you reserve H100 capacity
- SXM or PCIe: the SXM version has NVLink between the eight GPUs and much higher power.
- How the servers are interconnected (InfiniBand or RoCE, and at what speed per GPU) if you will train across nodes.
- Whether the price covers storage, egress and support, or only the GPU-hours.
- The uptime they commit to, and what you get back when they miss it.
Questions
- How much does it cost to rent a H100?
- Owners set their own price per GPU-hour, which depends on the number of servers, the term and the start date. Send a request with what you need and we come back with offers.
- What is the minimum term for reserved H100 capacity?
- Usually one month or more, set by each owner. Longer terms usually get a lower price per GPU-hour.
- How much memory does a H100 have?
- 80 GB HBM3 per GPU (NVIDIA Hopper).
- Can I rent a single H100 server?
- Yes, where the owner allows it: each listing shows how many servers it has and you can ask for one or for all of them.
Tell us what you need
Sign in to send a request. It opens a conversation with the owners and the Paralon team, and you get an email for every reply.
Sign in to requestFree, with Google, GitHub or an email code. No commitment.