New: connect Claude & other AIs to GenAIList over MCP โ research the catalog and contribute to the shared knowledge base. Learn how โ
Run FunSearch 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 โ | ~828.6 | yes | 636k | โ |
| AMD Instinct MI325X โ | ~550 | yes | 861k | โ |
| NVIDIA H200 โ | ~497.1 | yes | 457k | โ |
| AMD Instinct MI300X 192GB โ | ~485.8 | yes | 636k | ~$15,000 |
| NVIDIA H100 SXM5 80GB โ | ~347 | yes | 243k | ~$30,000 |
| Intel Gaudi 3 โ | ~317.1 | yes | 411k | โ |
| NVIDIA A100 SXM4 80GB โ | ~211.2 | yes | 243k | ~$15,000 |
| NVIDIA H100 PCIe 80GB โ | ~207.1 | yes | 243k | ~$25,000 |
| NVIDIA A100 PCIe 80GB โ | ~200.4 | yes | 243k | ~$12,000 |
| NVIDIA RTX PRO 6000 Blackwell โ | ~185.6 | yes | 299k | ~$8,500 |
| NVIDIA RTX 6000 Ada 48GB โ | ~99.4 | yes | 131k | ~$6,800 |
| NVIDIA L40S 48GB โ | ~89.5 | yes | 131k | ~$10,000 |
| AMD Radeon Pro W7900 48GB โ | ~79.2 | yes | 131k | ~$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 FunSearch?โถ
On GenAIList, 13 Datacenter GPU options can run FunSearch at Q4 โ the fastest is the NVIDIA B200 at ~828.6 tok/s.
How fast is FunSearch on Datacenter GPU hardware?โถ
Estimated generation speed ranges from ~79.2 to ~828.6 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.