🔌
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 ByT5-XXL?
Yes — here's how fast and up to what context, with real measured numbers.
ByT5-XXL on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 625k | 135.8-161.9 t/s | ~1.1-1.3 (slow) |
| Q8 | yes | 704k | 271.6-323.7 t/s | ~2.3-2.7 (slow) |
| Q4 | yes | 739k | 485-578.1 t/s | ~4-4.8 (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 ByT5-XXL?▶
Yes. At Q4 it generates ~485-578.1 tok/s and fits up to 739k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~4-4.8 tok/s.
How much context fits?▶
At Q4, up to about 739k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.