🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN ON DATACENTER GPU
Run CARP on Datacenter GPU hardware
13 options ranked by estimated speed at Q4 / 8k context.
| Hardware | Est tok/s | Fits 8k? | Max ctx | ~Price |
|---|---|---|---|---|
| NVIDIA B200 → | ~19,329 | yes | 2667k | — |
| AMD Instinct MI325X → | ~12,830.5 | yes | 3565k | — |
| NVIDIA H200 → | ~11,597.4 | yes | 1951k | — |
| AMD Instinct MI300X 192GB → | ~11,333.6 | yes | 2667k | ~$15,000 |
| NVIDIA H100 SXM5 80GB → | ~8,094 | yes | 1095k | ~$30,000 |
| Intel Gaudi 3 → | ~7,398.4 | yes | 1768k | — |
| NVIDIA A100 SXM4 80GB → | ~4,926.5 | yes | 1095k | ~$15,000 |
| NVIDIA H100 PCIe 80GB → | ~4,832.3 | yes | 1095k | ~$25,000 |
| NVIDIA A100 PCIe 80GB → | ~4,675.2 | yes | 1095k | ~$12,000 |
| NVIDIA RTX PRO 6000 Blackwell → | ~4,329.7 | yes | 1319k | ~$8,500 |
| NVIDIA RTX 6000 Ada 48GB → | ~2,319.5 | yes | 645k | ~$6,800 |
| NVIDIA L40S 48GB → | ~2,087.5 | yes | 645k | ~$10,000 |
| AMD Radeon Pro W7900 48GB → | ~1,847.6 | yes | 645k | ~$3,999 |
Estimate at Q4 (memory-bound). Click any option for its full per-hardware estimate and measured numbers.
FAQ
What Datacenter GPU hardware runs CARP?▶
On GenAIList, 13 Datacenter GPU options can run CARP at Q4 — the fastest is the NVIDIA B200 at ~19,329 tok/s.
How fast is CARP on Datacenter GPU hardware?▶
Estimated generation speed ranges from ~1,847.6 to ~19,329 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.