// RUN โ€” FITS

Can the NVIDIA B200 run OPT-66B?

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

OPT-66B on the NVIDIA B200

QuantFits?Max contextSpeedBeyond context
FP16 yes 132k 44.2-52.7 t/s ~0.2-0.3 (slow)
Q8 yes 333k 88.5-105.5 t/s ~0.4-0.5 (slow)
Q4 yes 422k 158-188.3 t/s ~0.8-0.9 (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 B200 run OPT-66B?โ–ถ

Yes. At Q4 it generates ~158-188.3 tok/s and fits up to 422k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~0.8-0.9 tok/s.

How much context fits?โ–ถ

At Q4, up to about 422k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.