// RUN โ€” FITS

Can the NVIDIA H200 run GPT-1?

Yes โ€” here's how fast and up to what context, with real measured numbers.

GPT-1 on the NVIDIA H200

QuantFits?Max contextSpeedBeyond context
FP16 yes 1953k 14974.4-17846.2 t/s ~124.8-148.7 (slow)
Q8 yes 1955k 29948.7-35692.3 t/s ~249.6-297.4 (slow)
Q4 yes 1955k 53479.9-63736.3 t/s ~445.7-531.1 (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 GPT-1?โ–ถ

Yes. At Q4 it generates ~53479.9-63736.3 tok/s and fits up to 1955k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~445.7-531.1 tok/s.

How much context fits?โ–ถ

At Q4, up to about 1955k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.