Which GPU should I buy?
A Llama-3-70B chat assistant, 70 requests a second, sized on three GPU options. H200 needs 32 GPUs; GB300 NVL72 and Vera Rubin need 12 each, and GB300 has the lowest monthly cost.
Capacity Planner / demo 2
01What you enter, what you get
Llama-3-70B serving live chat (realtime promise), chat and batch summaries: 70 requests a second. Three topologies: DGX H200 racks, GB300 NVL72 racks, Vera Rubin NVL144 racks.
H200: 32 GPUs, about $57,600 a month. GB300 NVL72: 12 GPUs, about $47,520 a month. Vera Rubin: 12 GPUs, about $69,120 a month (pre-launch estimate). Every speed promise met on all three. GB300 and Rubin come as 72-GPU racks, so 12 GPUs is a sixth of one rack.
02The Capacity Planner's report
Every run produces a report you can share: the workload, the recommended cluster, each traffic pattern against its speed promise, and the economics. This is the summary report from this demo, exactly as the Capacity Planner produced it.
03Watch
The demo in 36 seconds.
The real Capacity Planner screens, sped up.