// RUN โ€” FITS

Can the Intel Arc Pro B50 run DeepSeek Coder 6.7B?

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

DeepSeek Coder 6.7B on the Intel Arc Pro B50

QuantFits?Max contextSpeedBeyond context
FP16 no โ€” โ€” โ€”
Q8 yes 50k 19.4-24.1 t/s ~3.5-4.3 (slow)
Q4 yes 72k 34.6-43 t/s ~6.2-7.7 (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 Intel Arc Pro B50 run DeepSeek Coder 6.7B?โ–ถ

Yes. At Q4 it generates ~34.6-43 tok/s and fits up to 72k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~6.2-7.7 tok/s.

How much context fits?โ–ถ

At Q4, up to about 72k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.