New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the NVIDIA H200 run ALBERT-xxlarge?
Yes — here's how fast and up to what context, with real measured numbers.
ALBERT-xxlarge on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1949k | 7455.3-8885.1 t/s | ~62.1-74 (slow) |
| Q8 | yes | 1953k | 14910.6-17770.2 t/s | ~124.3-148.1 (slow) |
| Q4 | yes | 1954k | 26626.1-31732.5 t/s | ~221.9-264.4 (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 ALBERT-xxlarge?▶
Yes. At Q4 it generates ~26626.1-31732.5 tok/s and fits up to 1954k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~221.9-264.4 tok/s.
How much context fits?▶
At Q4, up to about 1954k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.