Powerful cloud native instances featuring full passthrough of one or more NVIDIA H100 GPUs.
The NVIDIA H100 is a flagship data center GPU based on the NVIDIA Hopper architecture, designed for artificial intelligence, high-performance computing (HPC), and large-scale data analytics. Featuring advanced Tensor Cores, high-bandwidth memory (HBM), and the Transformer Engine, it delivers exceptional performance for training and inference of large AI models. The H100 supports multi-instance GPU (MIG) technology, enabling efficient resource partitioning and workload isolation. Its industry-leading computational power, scalability, and efficiency make it a key platform for generative AI and enterprise-scale computing.
| Name | NVMe Storage | Memory | vCPUs | Transfer | Price | Regions |
|---|---|---|---|---|---|---|
G4-gpu-h100-240m-12c
| 4800 GiB | 240 GB | 12 vCPUs | 50 TB | $3.0137 /hr $2 200 /mo | 15 |
G4-gpu-h100-480m-24c
| 9600 GiB | 480 GB | 24 vCPUs | 60 TB | $5.8904 /hr $4 300 /mo | 15 |
G4-gpu-h100-960m-48c
| 19200 GiB | 960 GB | 48 vCPUs | 70 TB | $11.6438 /hr $8 500 /mo | 15 |
G4-gpu-h100-1920m-96c
| 38400 GiB | 1920 GB | 96 vCPUs | 80 TB | $23.2877 /hr $17 000 /mo | 15 |