posted by Spheron AI 3 months ago
$1.90 / gpu / hour - 512 H200s available from 02/08/2026 to 03/30/2027 (USA)
Unverified
Interconnect: RoCE v2
Cores per node: 192
RAM per node: 96 GB
NVME storage per node: 3200 GB
Location: USA
Cloud provider: Spheron AI
Cluster interface:
Minimium bookable GPUs: 32
Minimium bookable duration: 52 weeks
Additional details
Chassis
8-GPU H200
CPU
Intel Xeon 8558 ×2
Memory
96GB DDR5-5600 ECC REG ×24
Primary Storage
SSD M.2 NVMe PCIe 4.0, 960GB, 1 DWPD, TLC, TCG OPAL, 80mm ×2
(1.92TB total)
Secondary Storage
SSD 2.5” NVMe PCIe 4.0, 3.84TB, 1 DWPD, TLC, SED/TCG, 7mm ×4
Expandable to ×8 (15.36TB to 30.72TB total)
Network Interface Card (NIC)
MCX623106AS 100GbE 2-port QSFP56 Adapter ×2
RoCE v2 Adapter
NDR/EN 400G 1-port OSFP Adapter ×8
Interconnect
3.2TB RoCE v2
Total Storage
17.28TB to 32.64TB
(1.92TB + 15.36TB to 30.72TB)
Total GPU Units
8 × H200 GPUs
Total GPU Network Ports
8 × 400GbE RoCE v2
Total vCPU Cores
96 cores
Total GPU Memory
1,128GB (8 × 141GB HBM3e)
Total CPU Memory
2,304GB (24 × 96GB)
contact seller
seller email domain is spheron.ai