Mixed RP6000 and H200 GPU Clusters on OpenMetal

Q: Can I build a mixed GPU cluster with RP6000 and H200 servers?

Yes, OpenMetal builds mixed GPU clusters that combine RP6000 and H200 nodes on the same private network, letting you route each workload to the right GPU.

Explore GPU servers

A mixed cluster typically sends big-model and bandwidth-bound work (large-scale training, large-model inference) to H200 nodes (141GB HBM3e, 4.8 TB/s) and cost-efficient training and high-throughput inference to RP6000 nodes (96GB GDDR7, FP4). Every node is single-tenant bare metal with dual Intel Xeon 6530P, 1TB DDR5-6400, and a 6.4TB NVMe data drive.

Nodes connect over a private 40 Gbps mesh (4x 10 Gbps LACP-bonded), which carries gradients, parameter exchange, and dataset traffic from OpenMetal storage nodes; east-west traffic is not metered. GPU-memory pooling is within a node (one or two cards); across nodes, distributed jobs use data and pipeline parallelism over the private network rather than a shared GPU-memory fabric.

Clusters are built to order: you specify the node count, GPU mix, RAM (up to 2TB per node), and storage. The whole fleet is billed on fixed monthly pricing with included egress.