🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA H200 run LLaMA-65B?
Yes — here's how fast and up to what context, with real measured numbers.
LLaMA-65B on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | no | — | — | — |
| Q8 | yes | 192k | 53.7-64 t/s | ~0.4-0.5 (slow) |
| Q4 | yes | 280k | 96-114.4 t/s | ~0.8-1 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA H200 run LLaMA-65B?▶
Yes. At Q4 it generates ~96-114.4 tok/s and fits up to 280k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~0.8-1 tok/s.
How much context fits?▶
At Q4, up to about 280k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.