// RUN โ€” FITS

Can the NVIDIA B200 run IBM Granite 4.1?

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

IBM Granite 4.1 on the NVIDIA B200

QuantFits?Max contextSpeedBeyond context
FP16 yes 439k 97.3-116 t/s ~0.5-0.6 (slow)
Q8 yes 554k 194.7-232 t/s ~1-1.2 (slow)
Q4 yes 604k 347.6-414.3 t/s ~1.7-2.1 (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 B200 run IBM Granite 4.1?โ–ถ

Yes. At Q4 it generates ~347.6-414.3 tok/s and fits up to 604k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1.7-2.1 tok/s.

How much context fits?โ–ถ

At Q4, up to about 604k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.